Support Mixtral on macOS #1558
Merged
Support Mixtral on macOS #1558
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
A follow-up of my previous PR (mlc-ai#1529). This PR makes Mixtral work on Metal GPUs that macOS comes with. There are honestly no much change needed, except for that Metal doesn't support fp64 data types. A python script to run Mixtral: ```python from mlc_chat import ChatConfig, ChatModule, callback from mlc_chat.support import logging logging.enable_logging() MODEL = "HF://junrushao/Mixtral-8x7B-Instruct-v0.1-q4f16_1-MLC" NUM_GPU = 1 def main(): cm = ChatModule(MODEL, chat_config=ChatConfig( sliding_window_size=1024, tensor_parallel_shards=NUM_GPU, )) cm.generate("What is the meaning of life?", progress_callback=callback.StreamToStdout(callback_interval=2)) if __name__ == "__main__": main() ``` Quantization formats: - 3-bit (19.662 GB): ["HF://junrushao/Mixtral-8x7B-Instruct-v0.1-q3f16_1-MLC"](https://huggingface.co/junrushao/Mixtral-8x7B-Instruct-v0.1-q3f16_1-MLC) - 4-bit (24.466 GB): ["HF://junrushao/Mixtral-8x7B-Instruct-v0.1-q4f16_1-MLC"](https://huggingface.co/junrushao/Mixtral-8x7B-Instruct-v0.1-q4f16_1-MLC)
jinhongyii
approved these changes
Jan 8, 2024
Merged
|
NOTE: this may take a few extra days until 4 outstanding PRs in TVM get merged. For those who are curious, I have a working branch of TVM if you'd love to build it from source: https://github.com/junrushao/tvm/commits/mixtral-debug/ |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A follow-up of my previous PR (#1529).
This PR makes Mixtral work on Metal GPUs that macOS comes with. There's honestly no much change needed, except that Metal doesn't support fp64 data types.
A python script to run Mixtral:
Quantization formats: