Code

Features/vllm-base-pipeline#138

Code

Merged

p-ferreira merged 32 commits intostagingmacrocosm-os/prompting:stagingfrom features/vllm-testmacrocosm-os/prompting:features/vllm-testCopy head branch name to clipboardMar 18, 2024

Commits

Commits on Feb 28, 2024

• isolates llms into specific module

  steffencruz

  committed
  a6d2185

  Copy full SHA for a6d2185

• implement base and hf pipeline

  p-ferreira

  committed
  76284e0

  Copy full SHA for 76284e0

• adds base llm class

  p-ferreira

  committed
  21fa3ce

  Copy full SHA for 21fa3ce

Commits on Feb 29, 2024

• adds vllm model wrapper and pipeline + overall improvements

  p-ferreira

  committed
  b28ad5f

  Copy full SHA for b28ad5f

• drop-in replacement for hf zephyr

  p-ferreira

  committed
  f9de780

  Copy full SHA for f9de780

Commits on Mar 1, 2024

• Merge branch 'main' into features/vllm-test

  p-ferreira

  committed
  e76136e

  Copy full SHA for e76136e

• update pipeline param type on tasks

  steffencruz

  committed
  76aedc2

  Copy full SHA for 76aedc2

Commits on Mar 4, 2024

• improving variable name, removes unused commented code

  p-ferreira

  authored
  145b3cb

  Copy full SHA for 145b3cb

• fix broken tests

  p-ferreira

  committed
  5aaccd5

  Copy full SHA for 5aaccd5

• code declaration improvement on base_llm

  p-ferreira

  committed
  27c02bf

  Copy full SHA for 27c02bf

Commits on Mar 5, 2024

• unit test + overall adjustments

  p-ferreira

  committed
  7e035bf

  Copy full SHA for 7e035bf

• runs black on new changes

  p-ferreira

  committed
  195aa33

  Copy full SHA for 195aa33

• adds gpu calc utils functions + update vllm_llm

  p-ferreira

  committed
  ac3dee5

  Copy full SHA for ac3dee5

Commits on Mar 6, 2024

• update readme with vllm gpu device limitation

  p-ferreira

  committed
  40aac18

  Copy full SHA for 40aac18

• updates unit test

  p-ferreira

  committed
  4f4e9ac

  Copy full SHA for 4f4e9ac

• upgrade load_vllm to handle max-len vs kv cache exceptional scenarios

  p-ferreira

  committed
  5cd27ca

  Copy full SHA for 5cd27ca

• runs black on vllm file

  p-ferreira

  committed
  f3245cc

  Copy full SHA for f3245cc

• refactor vllm load second attempt approach

  p-ferreira

  committed
  d70dc3a

  Copy full SHA for d70dc3a

• adds extra cuda sync before mem_get_info

  p-ferreira

  committed
  7895c88

  Copy full SHA for 7895c88

• fix imports of vllm_llm

  p-ferreira

  committed
  b58ee03

  Copy full SHA for b58ee03

• adds tests for new vllm functionality

  p-ferreira

  committed
  9bc6060

  Copy full SHA for 9bc6060

Commits on Mar 7, 2024

• Update neurons/miners/zephyr/miner.py

  Show description for 75c8a4e
  

  p-ferreira

  and

  steffencruz

  authored
  75c8a4e

  Copy full SHA for 75c8a4e

• pr adjustments: file renaming + vllm requirements

  p-ferreira

  authored
  c8b9e28

  Copy full SHA for c8b9e28

• fix renaming imports

  p-ferreira

  committed
  7244789

  Copy full SHA for 7244789

• rollback on requirements

  p-ferreira

  authored
  fb3eb98

  Copy full SHA for fb3eb98

Commits on Mar 11, 2024

• Merge pull request #149 from opentensor/staging

  Show description for 1a4c69d
  p-ferreira

  authored
  1a4c69d

  Copy full SHA for 1a4c69d

Commits on Mar 12, 2024

• Merge branch 'main' into features/vllm-test

  p-ferreira

  committed
  3a0517b

  Copy full SHA for 3a0517b

• fix merging issues

  p-ferreira

  authored
  f5e241a

  Copy full SHA for f5e241a

Commits on Mar 13, 2024

• adjust cuda mock on unit test

  p-ferreira

  committed
  31dbfb9

  Copy full SHA for 31dbfb9

• Merge branch 'features/vllm-test' of https://github.com/opentensor/prompting into features/vllm-test

  p-ferreira

  committed
  1cb5a1e

  Copy full SHA for 1cb5a1e

• mock cuda device call on unit test

  p-ferreira

  committed
  12769d9

  Copy full SHA for 12769d9

• fix mock target on patch

  p-ferreira

  committed
  e3545ac

  Copy full SHA for e3545ac