Code

Add LoRA support for Gemma#3050

Code

Merged

WoosukKwon merged 13 commits intomainvllm-project/vllm:mainfrom gemma-loravllm-project/vllm:gemma-loraCopy head branch name to clipboardFeb 28, 2024

+82-7

Commits

Commits on Feb 27, 2024

• Add LoRA support for Gemma

  WoosukKwon

  committed
  3ea19ce

• Add test for Gemma + LoRA

  WoosukKwon

  committed
  fa04837

• Add punica kernel support for Gemma 7B

  WoosukKwon

  committed
  c55658f

• Add more sizes to punica test

  WoosukKwon

  committed
  08573e2

• Fix for memory usage

  WoosukKwon

  committed
  f8adea9

• Explicitly call GC for memory reuse

  WoosukKwon

  committed
  de2b481

• Empty cache

  WoosukKwon

  committed
  1237418

• Use cleanup

  WoosukKwon

  committed
  7afa08d

Commits on Feb 28, 2024

• Reduce max_model_len for test_llama_lora_warmup

  WoosukKwon

  committed
  d0f8bb3

• Ensure GC

  WoosukKwon

  committed
  e38361a

• GC for gemma

  WoosukKwon

  committed
  2358196

• Remove cleanup

  WoosukKwon

  committed
  ce07367

• Add --forked

  WoosukKwon

  committed
  071ac03