Skip to content
This repository has been archived by the owner on Oct 11, 2024. It is now read-only.

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
@mgoin
mgoin authored Jun 4, 2024
1 parent 061e5f7 commit 0257d9d
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@
[vLLM](https://github.com/vllm-project/vllm) is a fast and easy-to-use library for LLM inference that Neural Magic regularly contributes upstream improvements to. This fork, `nm-vllm` is our opinionated focus on incorporating the latest LLM optimizations like quantization and sparsity for enhanced performance.

## Installation
The [nm-vllm PyPi package](https://pypi.org/project/nm-vllm/) includes pre-compiled binaries for CUDA (version 12.1) kernels, streamlining the setup process. For other PyTorch or CUDA versions, please compile the package from source.
The [nm-vllm PyPi package](https://pypi.neuralmagic.com/simple/nm-vllm/index.html) includes pre-compiled binaries for CUDA (version 12.1) kernels, streamlining the setup process. For other PyTorch or CUDA versions, please compile the package from source.

Install it using pip:
```bash
Expand Down

8 comments on commit 0257d9d

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bigger_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 20.27275230837236 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2635.457800088407 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.0249983647596 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1563.249787418748 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.174333269804189 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3354.4312636296463 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0066501872083062 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4111.626233589819 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.164007999633388 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4268.108199624223 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7078786283674863 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 557.002981512763 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 388.18487202739817 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.979238649831835 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3592.664333006782 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.387000311179694 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 700.3100404533602 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 20.678907820364167 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2688.2580166473417 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.402917430053122 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1352.3792659069059 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.355993889685105 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 727.5937396384854 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 540.7100216582903 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.214209552028226 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3123.633032211641 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.94482759657841 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3583.8206923206512 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.46375928655739107 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.59012980858076 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.02466998054118 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.770945775744573 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2698.2634069071323 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.94721225333089 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 279.2886902687907 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.47155175236145 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839728261600236 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.1276677027035 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.39709434548215 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.2373630741039543 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1525.9828827880845 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7462446788727134 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 97.01180825345274 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4659640866439093 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 126.00290188300167 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.07219312832233 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9094401940657575 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.22722522854846 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.47486780370367165 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.52655523102855 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.40566024196859 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.771584959314922 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 880.3060447109399 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.49190862388145706 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03112563682436 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.99661811342723 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7954961958807603 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 233.41450546449883 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.766103207676531 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3860.255787868444 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.054038320000809 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4155.38927800083 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 48.17490110098406 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3131.3685715639635 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.452164744382777 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 757.2938641252517 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 563.000677545323 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7145988417999306 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.89784943399098 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7922008835819112 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 232.98611486564846 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.55629898686959 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3167.762569306177 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3957752670486614 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 441.450784716326 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9679674771361912 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 288.6253157742124 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 210.02313007249654 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.8729003942073446 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3969.7229040625284 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.103589743550874 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4157.141538441599 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.007614556240243 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1300.9898923112316 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.586984617386864 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1663.1540001301462 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.073536912517766 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4248.677133748903 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.6317471413035864 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 862.1271283694663 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3988855738657966 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 441.8551246025536 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.539861044967185 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1438.636566539651 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1037.7910488612886 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.494593456544024 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3844.7264432070842 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7946605117326093 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 233.30586652523922 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 49.98463972017833 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3249.001581811591 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.590307907788368 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1137.7326728990558 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 820.7970455697351 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.601393646344692 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 728.18117402481 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.204860665579496 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2106.6318865253343 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 28.82805592398578 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3718.8192141941654 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.767309172269687 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 879.7501923950593 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.491262686077927 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 843.8641491901305 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.477867374562016 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1418.9913923249574 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1023.4912195790581 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.302402522057358 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 711.0432962119004 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 461.726871643412 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.497423304325887 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3846.17815511918 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.855745983176633 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1480.6064575398273 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.299005988405791 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1362.312007665911 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 859.354101058364 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.3939034055780395 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3457.093476312068 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.656602842682746 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1385.3583695487569 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9094431945202156 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.22761528762803 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0443407491257277 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4188.854194958616 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.5556614640770485 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1126.753561351376 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 812.7696905455641 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7673294612278183 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3861.5126977585137 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.979314738841224 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 257.31091604935915 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.733542354831152 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 485.3605061280498 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4919127180478424 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03220788876666 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.99435407586492 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.231240331033986 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 689.0665139654558 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 512.1946054308377 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23914220248684964 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.088486323290454 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23940271756900844 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.122353283971094 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.97285636680128 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4105.024086267928 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.043706385952141 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4187.554384815938 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9834690318052302 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 289.97912225787815 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.27195582484086 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.229988530382094 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3125.66852041929 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.980491584060783 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3093.4834143438406 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839543592537457 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.1222226738311 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.3931081406494 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.781157999143534 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 361.55053988865944 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14.23388707632148 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3658.1089786146204 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.8968708481246166 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 376.59321025620017 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.00673677964333 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1560.8757813536329 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.770723482840023 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2698.1594716366812 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.881875840070942 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4043.402305956393 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.397743677853447 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 441.7066781209481 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23923793691131248 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.100931798470622 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.044576413845634 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4126.86770030281 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.062601728023291 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4226.270940719723 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.76563837400454 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2695.781878149563 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.621106215969444 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1665.3719040380138 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.240599107681863 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1705.638941999321 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.236687930982335 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1705.3847155138517 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.4482784169136833 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 756.0936625673964 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 561.5534595594352 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.492765940508571 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3843.7889274808967 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 29.35104516263643 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3786.2848259801 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.630574741561471 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4017.057708581298 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7632885481827683 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3857.3707618873377 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9086285468467634 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.12171109007924 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.960404356697984 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3587.8239196713816 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.766146386956784 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2179.599030304382 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.041641787316767 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4183.324022212057 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.87837507776493 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1544.1887601094409 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.8482295936065858 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 587.2392209296829 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 413.84571337588744 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.011747367829572 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 131.52715781784437 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.2553880277828078 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33.20044361176501 tokens/s

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smaller_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 73755.18542200007 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 54781.31541236267 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 66663.55898900019 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 66.39745048123163 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 64.54248457409597 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5274.142381000274 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 181.30165391202172 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 172.8570250006669 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.36751358898898 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.34859982848261 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6376.852755000016 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 108.15523370000392 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65.50477400003274 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.33136610131439 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 37.037428053801804 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6031.7559444999915 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 126.05798475333054 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 82.35389249989566 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.506726089476274 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.345374117580604 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1847.9448284997488 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 91.11867583997689 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.75746250012526 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.720778409994828 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.957975231980813 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6115.664882499914 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 105.67819980002848 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.5118680000869 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 34.6407318778397 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.16146819897536 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5123.928123499354 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.0871250799925 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 59.15235899919935 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.38484088215165 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30.957168229789495 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2053.7065545004225 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.75048137999693 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.944025999830046 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.684640059364112 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.797214167940783 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2412.450625499787 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 114.58843513332977 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.94715549966713 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17.97720822256229 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.207593782838163 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6051.014324999869 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 120.20469047333943 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 77.93934050005191 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.5328849296462 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.22912378618021 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11522.47239799999 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 189.9282906139882 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 149.82469449978453 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 87.57810831842728 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.81603877024982 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 57308.29728949993 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24429.24578628534 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 22884.999396499552 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 181.12438337329968 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 192.18140868332165 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12847.864813999877 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 202.18938731532884 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 156.56327750002674 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 100.64073281152855 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 103.13423978338325 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5772.151614000904 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.26393409336985 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 86.76324100088095 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 52.595730911464535 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 45.781826257653435 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12629.358229999525 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 162.56727568198158 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 128.46049300060258 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 112.88163107184694 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 101.52381771349468 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 55257.30873649991 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 34014.00603094267 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33341.29117949999 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 95.60821701314372 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 97.78219893172137 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1968.1786680002915 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 77.12485840001439 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 38.845241499984695 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.227909963066596 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.32405219681264 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7460.377201500023 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 155.0358747626657 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 133.64867150005466 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 55.98506018080652 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 55.65100405095286 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3654.1526315004376 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 142.80341375336016 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 93.23372050039325 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.38280171831792 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.700130815657285 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1929.135523500463 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 93.89699821668425 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.17522600000302 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.259917131621624 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.53502009480749 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2544.4823535003707 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.51974250535325 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 82.8322054999262 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.874461210297007 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.96535923873111 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 249719.7052050003 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 235743.81317350536 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 236612.42044949994 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.52977677315367 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.11.4 (main, Jun 7 2023, 10:57:56) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65.70908300783768 ms

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bigger_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 49.213588994327125 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3198.883284631263 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839747255220801 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.12822773593706 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.41062399845143 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4918760409597663 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.02251266730462 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.99837485244764 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.8626378313886933 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3959.203777173411 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.761079684724929 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3855.1066768430524 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7809510292251072 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.52363379926393 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.470951311212545 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 841.2236704576309 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.4036845152806885 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1078.5935860472973 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 778.041435978982 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.462871071246838 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1655.0866196310444 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7132456920229051 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.72193996297767 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.324682502155706 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1472.2087252802419 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.352757562912411 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.8584831786134 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.47463466097562035 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.46492628229548 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.35685569756319 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.98849224028599 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3094.515498996893 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.015582291885051 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4111.993715737031 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.4193371772900427 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1083.5537581114415 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 781.9249074702142 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.759818564120088 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3853.81402822309 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.55814615941132 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3998.443562968709 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.447207763609519 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 708.1370092692375 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0621612781433614 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4225.368458915747 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.761834382981533 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3855.880242556071 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.2387634487595077 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.039248338736 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.165344655663026 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4269.478272054602 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.974829160743685 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1296.7277908966792 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14.12970857648678 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3631.3351041571023 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4634018921850181 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.49565618018768 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 114.93293729972818 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.318957740479321 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 821.4645062623117 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.094193030926098 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4152.3210248650885 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.931667816314912 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3580.4386287929324 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.632599116158742 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2633.5780427511813 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.581184697902198 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 855.5540107272857 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7806214571520893 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.48078942977162 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 28.57029800374699 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3685.5684424833617 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9046591643650991 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.60569136746288 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9678426147811222 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 288.58808473338576 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 209.9992644131586 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.679032646699512 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 547.5952299741876 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 381.6262109132343 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.876289763050417 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4040.536648444864 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.9619796079652974 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 255.05734903548864 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.25481106121001057 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33.12543795730137 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.997761339723258 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2469.7089741640234 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.989215434079231 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2078.5980064303 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.55358912166087 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1660.9832929079564 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7395326517053977 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 96.1392447217017 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.107263206472926 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1696.9721084207401 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.606442085198648 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3045.2310289906254 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9467798169630782 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 279.1611849642868 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.37189128965008 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.8025928571705327 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 572.7390302373648 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 403.618566649054 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.49190699173934305 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03069419637794 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.00605147113679 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.8411493435819906 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 369.34941466565874 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.573552263492296 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2154.5617942539984 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.124881221520817 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1307.133610287732 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 822.952302584804 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.293730713737247 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 708.3652105544281 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 459.2171221206701 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.266754671283298 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2504.6781072668286 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.2759927449989 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1465.879056849857 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23878307423102732 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.04179965003355 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9045774035685509 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.59506246391162 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.2369883881320245 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 680.8084904571632 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.1939069393747825 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1505.4991383853337 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.925561080202264 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3578.869197611982 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.456196401408729 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1489.3055321831348 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.486907451278414 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 843.2979686661938 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.34926290449463 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3141.054914679807 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.072710873860755 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1309.4524136018981 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.005004353236156 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1690.3252829603503 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0408186297413824 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4181.637372340092 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.042539237561269 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4185.16289776304 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.4762337816953375 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3835.307930009708 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.4473978677637893 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 755.8217255085985 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 561.8638128897397 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.591458342120607 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1246.889584475679 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.643798222277785 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2638.814296808201 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.348690127667196 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.32971659673547 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.626820789987049 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2630.8763285663445 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23851498352108885 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.00694785774155 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.040536369114373 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4181.059020315351 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.1088363025017385 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3323.807501597713 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.3525535679622855 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1379.2807001515687 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 994.9270064814698 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.917225326439533 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3576.72690889496 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.3459698126173945 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 724.4980373312545 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 538.3969440359418 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.005403330671799 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4109.071424546517 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.442837900940011 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 754.4134861543005 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 559.865875723438 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9828862956667367 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 289.80730056498953 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.14617805670846 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839008104601105 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.10643363353176 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.39466795402404 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4655465333779219 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.88998991250179 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.96638791440957 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.740448880851695 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 356.2583545107203 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.485149220812231 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3839.8815502766747 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7802903363439242 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.43774372471015 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.923251594754908 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3237.035515643605 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.246746972456054 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1345.751648101599 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 970.2004886687168 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3554389722728466 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 436.20706639547006 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.2205986053258457 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 685.7800652874299 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 509.7073214720734 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.049221313894132 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4150.4518467414855 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.484183431870546 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3839.3861005495905 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7918179439755586 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1456.0580904866147 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 29.08916839524036 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3752.5027229860066 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.37851486335967 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3144.8284173733973 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.07240936566586 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4246.366790249347 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.9175350653366 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3114.639779246879 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.0071697311074252 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.93206504396528 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9055297218576845 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.71886384149899 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.92645773378011 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4093.099637581488 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.6751287693341737 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 477.76674001344253 tokens/s

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smaller_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1975.7934499998555 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 95.33599586665939 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 59.25566899941259 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.613003521163858 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.903863959951174 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1990.530694500194 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.04017088661931 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.069916500215186 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.352486622843355 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.455696360151357 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65623.97806299999 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 42548.72657927333 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 42052.41047800007 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 100.42581577642753 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 102.73442258920154 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5165.230333999716 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.18472634663583 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 60.3192204998777 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.74831494379642 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.263469881805783 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 71233.34427600002 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31836.63932091001 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31021.99345149984 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 193.19795882911964 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.57356740026793 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6422.19940900003 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 109.88173289333115 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.78409450004119 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.57983124137713 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 37.267313983479774 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6109.015791500269 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.68514260334541 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 83.57903599994643 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.146640564830165 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.725523279960974 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 78529.7785214998 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58721.69160256535 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 71508.66835149986 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.09258718482668 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65.70630902656838 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6104.897983000001 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 126.89376464333085 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 85.76336650003213 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.06915336348656 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.874942475419786 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 260933.7399484998 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 246348.652753672 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 247753.58967099988 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 71.52547558982853 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.37765993345238 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2068.365681999694 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.17974005337844 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.95567000013034 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.835997486112298 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.943342617564616 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18863.58120349996 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1483.9681253440042 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 265.1577640008327 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.88222256506052 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 145.82300105146194 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6040.10625599949 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 119.45821917999032 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 91.84418099994218 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 54.64609627110425 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.86997792578394 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2505.3557870000986 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.1015805199798 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 81.43590049985505 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.655124752041274 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.882650920331617 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15549.75253199973 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.969658120659 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 168.86559050044525 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.20656284526063 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.60245925788612 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5455.755123000927 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 184.0125064293631 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 172.26980849954998 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.34378620451925 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 38.65310148491302 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2630.551425500016 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 120.0175972480065 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 83.42846149980687 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.839985249298476 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17.839909343441953 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3697.770496000885 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 144.23739775337404 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 95.34760749920679 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.870631533442317 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.040367541896167 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1896.827923500041 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.33098896669617 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 57.04740800001673 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.077754018439807 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.319771208280605 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6178.988775500102 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 105.24486398669069 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.42057300013948 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.014338630567 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.53110167117665 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17629.28239049961 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 228.15737318399079 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 183.52132100017116 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 146.45657947859132 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 152.18902241546186 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8023.916034499962 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.39932492533202 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.8959225000308 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.86249245694424 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.9.17 (main, Jun 7 2023, 12:29:40) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.52022299866027 ms

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bigger_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.943368534667666 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3583.44571340959 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4918915704450231 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.0266177314374 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 123.45494635029189 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9826167633295881 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 289.72782805694015 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.0880021970583 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0064286052670615 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4111.172212192209 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.4239596101105048 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1085.018560847918 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 783.0070621182508 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.2390179318220308 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.072331136864 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.753773551241663 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3847.6178900227046 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0616669744411835 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4224.355630629984 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.974537772997369 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1296.6899104896581 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.4724262055308905 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3833.3546434373466 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.994903359991445 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2079.3374367988877 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.351354237432726 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.67605086625434 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9676517331493559 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 288.5311682846978 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 209.9546220471929 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.4632820966125592 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.4639894185639 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 114.90631414915768 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.296389575900825 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1644.2653224335536 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9463931920480187 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 279.04718731933184 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.2947437234323 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.189624661214953 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1503.4806180582475 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839274771243902 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.11429638838354 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.37746617904256 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.97824727090582 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1688.5860726088783 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.348134209115219 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.25744718497845 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.6732141128359506 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 545.697588856016 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 380.30595118255843 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.292709678242438 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 708.0498875660179 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 459.1044137562164 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.4778575776457314 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3836.14093733226 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.8738981356156295 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4039.3097435708182 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.922212118651605 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3236.5494981967445 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.677905212124062 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 478.127677576128 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.079431998470541 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4144.748615215388 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23873526044230056 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.035583857499073 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.474500680788467 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.42950995962337 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.32880917838577 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.703917434225154 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3100.7546332246347 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.983902428973314 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.1069108575449 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.38189864341632 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.301599416228 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1469.2079241096399 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.0071109926798198 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.9244290483766 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.114316545877382 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1303.7857702230835 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 820.918608888125 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.617120343786242 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1250.2256446922115 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.8123095573987293 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 575.8263238786699 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 405.7942329971495 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.49192414177280086 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.03522763622217 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.98734867681915 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.476136303821353 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3835.2579238603544 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.2305316049716 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3125.738577041336 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 28.64203610152703 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3694.822657096987 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.531639075371887 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3991.631242370575 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.932130351364084 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3580.5575003005692 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.46679590928699 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 710.6834682073087 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.7425793261528315 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 356.53531239986813 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.4431478452613544 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 754.5092052259132 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 560.516751251241 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7099686308531521 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.29592201090978 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.3463000646917678 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 724.600027978543 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 538.4414524458119 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.8518845015708725 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 370.7449852042134 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.33352869168675 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 823.3587299192775 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.782248678891113 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.69232825584467 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23873308167562915 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.03530061783179 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.938336621783662 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2461.983760831876 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.594077559366856 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 857.2300827176913 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.037259458604506 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4174.344630680633 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.44041880548091 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1653.6272223562592 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7810795775329642 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.54034507928534 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.269079160485647 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1352.8284951662968 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 975.4646658015547 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.02178913768373 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1691.4162939494424 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.4471997051750036 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 755.7605276168457 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 561.9684144305073 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.461586665781842 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 840.0062665516394 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 29.053164064251956 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3747.8581642885024 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7427211329198625 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 96.55374727958213 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7936311552796704 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1456.7543636273933 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.049905659077816 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4151.153300554762 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14.19596395396433 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3648.3627361688327 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.063764214481623 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1308.2893478826109 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.620021662923268 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2627.697328716403 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.398134615478988 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3147.3593653967896 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.356235526617929 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 436.3106184603308 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7535171722622485 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3847.3551015688045 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.012987771020331 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4110.66272653343 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9051560820464887 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.67029066604353 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.253048858394748 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 682.8963515913173 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.379287036358612 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3015.928027690261 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.16903815451819 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4273.264108381144 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.439970061526557 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1487.1961079984524 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9065141662661236 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.84684161459607 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.623865152296648 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2629.4943906078206 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.036649740180936 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4173.095317630738 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9058331780606509 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.75831314788462 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.347115491581846 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1377.557428127371 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 993.7824802225482 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.47216861120863 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 841.381919457122 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0335880663929986 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4166.8219480392545 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.853977058785304 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3560.472104107823 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.628188348427338 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2631.5157441906863 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.2546884979044676 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33.10950472758079 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.220844467891792 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 685.8559942041292 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 509.75783353009604 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.58426796450528 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2155.9548353856862 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 49.200862371424066 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3198.056054142564 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.027711446841614 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3099.574776642568 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.46495491655968735 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.73000883662692 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.81647615678105 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.918728904256996 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4091.113328394048 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0687410545471723 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4238.850420767156 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.780460028149053 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.4598036593769 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.8604598938691095 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3956.9713912158372 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.168053973040166 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2491.8470164952214 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.962804862093478 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 255.16463207215213 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.401500138536137 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1077.9013789007165 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 777.5194370001083 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.277992243720124 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1466.138991683616 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.108292815428348 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3323.553388781678 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.750132482801772 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3843.8857948718164 tokens/s

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smaller_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1997.095519499453 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79.49410707335119 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.92237950044364 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.423041893780688 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.522508858599581 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3708.8484034998146 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 144.70396740337796 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 96.86096099994757 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.751408917969854 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.16263272998981 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 70348.65259299931 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31540.73703901402 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30937.1006654992 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 193.07305639892365 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.23880160253765 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6097.933693500181 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.86066675333373 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 84.7730484999829 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.20517006487294 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.686507297140324 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6452.393932000006 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 110.7267989666669 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 70.27968650004368 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.68642417315979 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 37.35562801671746 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6124.660605999907 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 128.36626168998959 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 84.43713399992703 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.95552879690462 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.76947833644865 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1887.9630804999579 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 91.94640251327655 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 60.55626250008572 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.012353670994822 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.249874644191436 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 79988.87884600004 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 59738.638552652 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 73359.15299499992 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.03593499119127 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 66.07576343076235 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6063.049027000488 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 119.42886400928546 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.7157725003126 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 54.8214165161257 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 48.03518404416397 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5198.608692498965 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 90.76383455999651 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 62.78620149987546 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.83553935022538 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.3345803907413 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1966.553798500172 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 95.29166702000414 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.26092300048913 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.558316077371831 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.834920917726885 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19222.069294000903 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1812.5878313566814 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 295.03822549941106 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 159.2759532426503 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 147.19713947191278 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 259086.0955725002 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 244112.833214076 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 245980.72280350016 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 70.04524546886341 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.15701582574867 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2071.589972500533 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 81.63230001999182 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.06669199973112 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.828980113139927 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.936821029889687 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2636.5531265000755 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 120.24517320798866 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 81.84799849959745 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.85569717316406 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17.877478772904112 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5440.578917499806 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 182.5354573026731 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 171.29180650044873 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.02768010298062 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 38.25011724984416 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17301.371945000028 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 224.18166553200354 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 181.0413339999286 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 142.13134217250737 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 147.33623283996045 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2525.048247500308 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.72952551065949 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.95160250013578 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.661041977349463 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.853015171401005 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15602.865738499986 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 210.0430408546581 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 172.34792499994 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 126.3320489613039 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 131.8927493348655 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8026.447241000028 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.6388383813328 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 129.0641964999395 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.9432166560783 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.63410415277631 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6170.664217999956 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 105.15708289333816 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.19553099990117 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.07245940430774 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.50399300944225 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65001.06689899997 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 42183.08303530666 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 42266.78036499993 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 100.48058928235726 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.8.17 (default, Jun 7 2023, 12:29:39) \n[GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 102.91195303058437 ms

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

bigger_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.42671716209927 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3151.0465139108055 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.545508303117494 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3090.458039702637 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.466009498581447 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3830.0628727722824 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7424052512599771 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 96.51268266379704 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.86620396508522 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3962.8590642123504 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.0081037950846143 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 131.05349336099988 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.680989182632066 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 548.2333293858017 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 382.08435857444823 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.7584788134402154 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3852.440783776221 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.483007455729199 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3838.7828247890793 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.1978475153297268 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1507.3565919109121 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0612108298520444 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4223.420990366839 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.933983711299275 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3581.0338138039137 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.46555601470405744 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.89255378950651 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.96879041924346 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.9660319512811197 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 255.58415366654555 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.7123686803007651 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 92.60792843909947 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.713231601446406 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2172.7201081880326 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.023458474382062 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4116.034197357998 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.614862861524769 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 859.93217199822 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.471101982955795 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3832.6753172563226 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0388046754009195 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4177.510779896484 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9050104897357941 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.65136366565324 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 29.20419758923193 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3767.341489010919 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.405732839956567 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2522.7452691943536 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.19013216848243 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2104.717181902716 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.918724642049467 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3577.112233006713 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.757715873044926 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3851.6587698710487 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.922665017019824 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4092.124909374095 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.144998410625755 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1318.849793381348 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.666397862169312 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1668.3158610410053 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.167162229085349 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4271.3412848124835 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.307572556302243 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1365.0266673666176 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 984.5847309090411 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.378476665053682 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1479.2019664569787 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.134851997346582 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1310.2932494391584 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 825.0986897985175 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.339519531917528 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 824.1375391492786 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.050283869164276 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4151.540965893382 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.352851151012647 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1475.8706496316443 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.8029581409673723 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - synthetic\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1460.3359261314708 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.924585530556524 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3578.6184813530267 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23871924520041385 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.0335018760538 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9828563033501835 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 289.79845723047947 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.1397045151036 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23883869923578863 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.04903090065252 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.2952469755363745 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 708.8334726316468 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 459.03715379007207 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.480756088088218 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 712.4982914514684 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.7459372961862396 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 356.97184850421115 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.357209648889467 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 436.4372543556307 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 49.52506453539982 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3219.1291948009884 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.6075187617996125 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 858.9774390339496 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.644993940393672 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2639.373366770465 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.448638330305748 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 756.2048134205564 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 562.2008309359856 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.649954137523288 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - 2:4 Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2641.6925565403885 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0056080209869043 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4109.490835002167 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 8.09313069780076 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4151.77604797179 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.906520610821843 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.84767940683957 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 28.69163175214443 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3701.2204960266313 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 23.350009520894893 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3012.1512281954415 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 15.566867236976975 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4000.6848799030827 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.3487816539442847 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 725.3664089154337 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 539.0422578713415 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9465912147530647 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 279.1055749739903 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 204.33117961659656 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0379347491809767 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4175.728301071821 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.25478761307773906 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 33.122389700106076 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.47461430245673736 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 125.45954471141395 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 99.35259398094368 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.446877653747609 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1092.2810596960799 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 787.8987836858076 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9677523000048635 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 288.5611549744502 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 209.97644237305525 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.221694466699758 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 686.1184965026639 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 509.9558988491067 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.23887545772215793 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.05380950388053 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.49189760815587785 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.02821373992475 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.03979658347329 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3566454527867053 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 436.3639088622717 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.4500754013323487 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1093.294393928208 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 789.0161439328352 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9059235167576485 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 117.7700571784943 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.444727823991582 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 754.9971447905737 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 560.3348768959693 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.867541009597012 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 512,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4036.0485379232673 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.527574977014785 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1498.584747011922 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839651097045238 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.1253924800779 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.34623044163428 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.49190176948969183 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 130.02931374690513 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.97852145292386 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.8161682353541437 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 577.0523441978956 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 406.6473325687342 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.778955382647688 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 64\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2571.2641997441992 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0383270965183287 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4176.532220766056 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 10.026286686579304 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1303.4172692553095 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.15377140945897 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3115.836511820207 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.782695497411685 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.75041466351905 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 9.609307023209363 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1249.209913017217 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14.179570276086457 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 256,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3644.1495609542194 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.652970270936615 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Sparse (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2643.102779879124 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7827275382188095 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.75457996844523 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.756812665145866 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 1024,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3850.7329817745126 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.0710419324566574 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2048,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4243.564919603691 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2.8501057587319583 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 4\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 370.51374863515457 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.683459835626144 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 478.8497786313987 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3.3514088897648815 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 16,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 435.6831556694346 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.46341443369488344 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 122.4989714029055 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 114.93913727456297 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.630734062614002 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 32,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 861.9954281398203 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.021273086151755 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - 2:4 Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"semi_structured_sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1626.382750599864 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5.256871226644719 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Dense (synthetic)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 683.3932594638135 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 25.495756053004822 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Dense (synthetic)\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1657.2241434453135 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 4.3726228329051136 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1385.6404495193015 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 999.4795517360079 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7.145913031722781 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3341.143097112303 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 26.133532158192033 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 64,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1698.6795902824822 tokens/s
{"name": "request_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.468251073424685 prompts/s
{"name": "token_throughput", "description": "VLLM Engine prefill throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 128,\n \"output-len\": 1,\n \"num-prompts\": 1,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3156.4043884717844 tokens/s
{"name": "request_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6.960648624268822 prompts/s
{"name": "token_throughput", "description": "VLLM Engine throughput - Dense (with dataset)\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"dataset\": \"sharegpt\",\n \"output-len\": 128,\n \"num-prompts\": 1000\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3254.5208707631305 tokens/s
{"name": "request_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1.7825348332403679 prompts/s
{"name": "token_throughput", "description": "VLLM Engine decode throughput - Sparse (synthetic)\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax_model_len - 4096\nbenchmark_throughput {\n \"use-all-available-gpus_\": \"\",\n \"input-len\": 2,\n \"output-len\": 128,\n \"num-prompts\": 8,\n \"sparsity\": \"sparse_w16a16\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 231.7295283212478 tokens/s
{"name": "request_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 0.9839127478595323 prompts/s
{"name": "input_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 290.10995341554263 tokens/s
{"name": "output_throughput", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 212.38740575295864 tokens/s

This comment was automatically generated by workflow using github-action-benchmark.

@github-actions
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

smaller_is_better

Benchmark suite Current: 0257d9d Previous: fdd9a31 Ratio
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 77863.63307100009 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58281.66058818401 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 71254.30481450008 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.18448559680624 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 65.57614076542656 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6169.880697000053 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 106.46133026002947 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 69.49139850007668 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 34.919925371456834 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 35.40104160368872 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16823.780649499895 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 222.3978496020136 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 174.9299864995919 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 135.8074422525249 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 142.19356786475004 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17811.0223725007 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1061.8634171593512 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 227.86593749970052 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 155.41090960013378 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 143.09971559737025 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 3699.151089499537 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 142.86275332994893 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 93.76811899983295 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.791819456090682 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 24.067366447461946 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6002.976175000185 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 118.60854235867495 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.63529949960503 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 54.369098626418904 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 47.3331707284874 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2522.696570500102 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 115.05474123196716 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 80.29174350031099 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 18.53805063470437 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 16.71867678207576 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5440.339908500391 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 183.3716414972999 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 170.48778099979245 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.94271573698874 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 38.24281609001152 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6110.254696499965 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 129.08200832666654 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 90.9221575001311 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.88085330292909 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.81224940285663 ms
{"name": "median_request_latency", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 5154.448082000272 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 89.23465427997144 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 61.30725599996367 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.775983947753744 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - 2:4 Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned2.4\nmax-model-len - 4096\nsparsity - semi_structured_sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 31.273593288596942 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 62147.10263950019 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39766.26892405266 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39408.58177550001 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 98.89464387296319 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 101.30746616648189 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6096.493981000094 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.65361824999916 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 84.98213850020875 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 40.331489254710824 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 39.851920693577306 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 7975.029886000016 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 157.9884472533413 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 135.43239950001862 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.62101486594874 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 58.21389066095032 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1992.651213499812 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 78.69765539995569 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.1947960001271 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.402564449955385 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.479954827523413 ms
{"name": "median_request_latency", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 69047.98245550046 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 30681.225544506644 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 29819.8609330002 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 191.35947444170938 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Sparse\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-pruned50\nmax-model-len - 4096\nsparsity - sparse_w16a16\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 201.84431503164717 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2616.2008219998825 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 120.6278214760144 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 84.79476649972639 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 19.69699261167778 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"750,2.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 17.6767269947816 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1964.6460755002408 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 95.28143891663603 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 60.55193900010636 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.63106773922918 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.841530274834774 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 2074.2727599999853 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 81.70022990000386 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 41.03294449987516 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.826978801172709 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 11.95713960298148 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 258395.27599550024 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 243924.9034436213 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 245393.95490599985 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 69.57896753179935 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - NousResearch/Llama-2-7b-chat-hf\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 67.31930061260277 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 6430.6718875000115 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 110.98738446666934 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 68.75269950000984 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 36.660171282886196 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - teknium/OpenHermes-2.5-Mistral-7B\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"150,0.5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 37.35048996213381 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 14931.094657999893 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 208.00185868199515 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 169.5139149996976 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 121.07384939298134 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"1500,5\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 127.14083376761975 ms
{"name": "median_request_latency", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 1884.3689440000162 ms
{"name": "mean_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 91.87971182333058 ms
{"name": "median_ttft_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 57.980469000540324 ms
{"name": "mean_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 13.001923244417402 ms
{"name": "median_tpot_ms", "description": "VLLM Serving - Dense\nmodel - neuralmagic/OpenHermes-2.5-Mistral-7B-marlin\nmax-model-len - 4096\nsparsity - None\nbenchmark_serving {\n \"nr-qps-pair_\": \"300,1\",\n \"dataset\": \"sharegpt\"\n}", "gpu_description": "NVIDIA A10G x 1", "vllm_version": "0.5.0", "python_version": "3.10.12 (main, Jun 7 2023, 13:38:31) [GCC 9.4.0]", "torch_version": "2.3.0+cu121"} 12.251956720453101 ms

This comment was automatically generated by workflow using github-action-benchmark.

Please sign in to comment.