Code

Implement LLaMA#9

Code

Merged

zhuohan123 merged 16 commits intomainvllm-project/vllm:mainfrom llamavllm-project/vllm:llamaCopy head branch name to clipboardMar 30, 2023

+500-35

Commits

Commits on Mar 25, 2023

• Add sentencepiece and Ray to dependency

  WoosukKwon

  committed
  1c80c4a

  Copy full SHA for 1c80c4a

Commits on Mar 26, 2023

• Implement LLaMA

  WoosukKwon

  committed
  6ed7057

  Copy full SHA for 6ed7057

• Implement LLaMA memory analyzer

  WoosukKwon

  committed
  bd8664c

  Copy full SHA for bd8664c

• [WIP] Add LLaMA model loader

  WoosukKwon

  committed
  4ea2216

  Copy full SHA for 4ea2216

• Fix README

  WoosukKwon

  committed
  742ac4a

  Copy full SHA for 742ac4a

• Minor

  WoosukKwon

  committed
  f3d5e78

  Copy full SHA for f3d5e78

• Minor

  WoosukKwon

  committed
  d2e08a2

  Copy full SHA for d2e08a2

Commits on Mar 28, 2023

• [WIP] Incorrect TP implementation

  zhuohan123

  committed
  48b6dd1

  Copy full SHA for 48b6dd1

• fix lm_head

  zhuohan123

  committed
  9144167

  Copy full SHA for 9144167

Commits on Mar 29, 2023

• Merge branch 'main' into llama

  zhuohan123

  committed
  20013d3

  Copy full SHA for 20013d3

• fix merge error

  zhuohan123

  committed
  fcb4f95

  Copy full SHA for fcb4f95

• fix distributed execution

  zhuohan123

  committed
  91ec0c5

  Copy full SHA for 91ec0c5

• Fix a bug in memory analyzer when using TP

  WoosukKwon

  committed
  7e72747

  Copy full SHA for 7e72747

Commits on Mar 30, 2023

• Fix the shape of q and k

  zhuohan123

  committed
  dd38548

  Copy full SHA for dd38548

• fix memory analyzer

  zhuohan123

  committed
  ffd9644

  Copy full SHA for ffd9644

• Merge branch 'main' into llama

  zhuohan123

  committed
  56b46a1

  Copy full SHA for 56b46a1