Commits

Breadcrumbs

History for

onmain

User selector

All users

Datepicker

All time

Commit History

Commits on Jun 17, 2023

• Change the name to vLLM (vllm-project#150)

  WoosukKwon

  authored
  0b98ba1

  Copy full SHA for 0b98ba1

Commits on Jun 7, 2023

• Support FP32 (vllm-project#141)

  WoosukKwon

  authored
  e38074b

  Copy full SHA for e38074b

Commits on May 27, 2023

• Improve setup script & Add a guard for bfloat16 kernels (vllm-project#130)

  WoosukKwon

  authored
  d721168

  Copy full SHA for d721168

Commits on May 15, 2023

• Add copyright headers to source files adapted from FT (vllm-project#104)

  WoosukKwon

  authored
  667ba39

  Copy full SHA for 667ba39

Commits on May 4, 2023

• Fix a bug in attention kernel (vllm-project#68)

  WoosukKwon

  authored
  130d5fd

  Copy full SHA for 130d5fd

Commits on May 3, 2023

• Support bfloat16 data type (vllm-project#54)

  WoosukKwon

  authored
  e070829

  Copy full SHA for e070829

• Refactor attention kernels (vllm-project#53)

  WoosukKwon

  authored
  436e523

  Copy full SHA for 436e523

Commits on Apr 28, 2023

• Add support for GPT-NeoX (Pythia) (vllm-project#50)

  WoosukKwon

  authored
  a96d63c

  Copy full SHA for a96d63c

Commits on Apr 15, 2023

• Support various block sizes & Change default block size to 16 (vllm-project#38)

  WoosukKwon

  authored
  0f4b321

  Copy full SHA for 0f4b321

Commits on Apr 11, 2023

• Memcpy kernel for flash attention (vllm-project#29)

  Show description for e3cec88
  suquark

  authored
  e3cec88

  Copy full SHA for e3cec88

Commits on Apr 10, 2023

• Support block size 32 (vllm-project#35)

  WoosukKwon

  authored
  b9926f7

  Copy full SHA for b9926f7

Commits on Apr 8, 2023

• Add query stride to multi_query_cached_kv_attention & Add kernel benchmark script (vllm-project#27)

  Show description for c267b1a
  WoosukKwon

  authored
  c267b1a

  Copy full SHA for c267b1a

• Implement block copy kernel to optimize beam search (vllm-project#32)

  WoosukKwon

  authored
  0f40557

  Copy full SHA for 0f40557

Commits on Apr 5, 2023

• Basic attention kernel that supports cached KV + (multi-)prompts (vllm-project#24)

  suquark

  authored
  21b3671

  Copy full SHA for 21b3671

Commits on Apr 2, 2023

• Optimize data movement (vllm-project#20)

  WoosukKwon

  authored
  897cb2a

  Copy full SHA for 897cb2a

Commits on Mar 31, 2023

• Add custom kernel for RMS normalization (vllm-project#16)

  WoosukKwon

  authored
  09e9245

  Copy full SHA for 09e9245

Commits on Mar 30, 2023

• Implement custom kernel for LLaMA rotary embedding (vllm-project#14)

  WoosukKwon

  authored
  88c0268

  Copy full SHA for 88c0268

Commits on Mar 13, 2023

• Add miscellaneous updates (vllm-project#8)

  WoosukKwon

  authored
  cfae35b

  Copy full SHA for cfae35b

Commits on Mar 10, 2023

• Support beam search & parallel generation (vllm-project#7)

  WoosukKwon

  authored
  1a7eb7d

  Copy full SHA for 1a7eb7d

Commits on Mar 1, 2023

• Implement single_query_cached_kv_attention kernel (vllm-project#3)

  WoosukKwon

  authored
  0deacbc

  Copy full SHA for 0deacbc

Commits on Feb 18, 2023

• Add reshape_and_cache op

  WoosukKwon

  committed
  c413c41

  Copy full SHA for c413c41

Commits on Feb 16, 2023

• cache_kernel -> cache_kernels

  WoosukKwon

  committed
  ffad4e1

  Copy full SHA for ffad4e1

• Remove redundant fn

  WoosukKwon

  committed
  6d2f74e

  Copy full SHA for 6d2f74e

• Implement cache ops

  WoosukKwon

  committed
  6f058c7

  Copy full SHA for 6f058c7

Loading