Commits

Branch selector

 main

User selector

WoosukKwon

Datepicker

All time

Commit History

Commits on Jan 15, 2024

• Announce the second vLLM meetup (vllm-project#2444)

  WoosukKwon

  authored
  2a18da2

Commits on Jan 12, 2024

• [Minor] Fix err msg (vllm-project#2431)

  WoosukKwon

  authored
  35c4bc2

• Rename phi_1_5 -> phi (vllm-project#2385)

  WoosukKwon

  authored
  50376fa

Commits on Jan 8, 2024

• [Minor] Remove unused code in attention (vllm-project#2384)

  WoosukKwon

  authored
  28c3f12

• Fix eager mode performance (vllm-project#2377)

  WoosukKwon

  authored
  c884819

Commits on Jan 4, 2024

• Bump up to v0.2.7 (vllm-project#2337)

  WoosukKwon

  authored
  2e0b6e7

• Revert the changes in test_cache (vllm-project#2335)

  WoosukKwon

  authored
  9417671

Commits on Jan 3, 2024

• Enable CUDA graph for GPTQ & SqueezeLLM (vllm-project#2318)

  WoosukKwon

  authored
  6ef00b0

Commits on Dec 21, 2023

• [BugFix] Fix recovery logic for sequence group (vllm-project#2186)

  WoosukKwon

  authored
  a1b9cb2

• Disable Ray usage stats collection (vllm-project#2206)

  WoosukKwon

  authored
  3a4fd5c

Commits on Dec 20, 2023

• [BugFix] Fix weight loading for Mixtral with TP (vllm-project#2208)

  WoosukKwon

  authored
  ba4f826

Commits on Dec 19, 2023

• Add warning on CUDA graph memory usage (vllm-project#2182)

  WoosukKwon

  authored
  21d5daa

Commits on Dec 18, 2023

• [BugFix] Fix RoPE kernel on long sequences(vllm-project#2164)

  WoosukKwon

  authored
  76a7983

• [BugFix] Raise error when max_model_len is larger than KV cache (vllm-project#2163)

  WoosukKwon

  authored
  8041b73

Commits on Dec 17, 2023

• Bump up to v0.2.6 (vllm-project#2157)

  WoosukKwon

  authored
  671af2b

• Disable CUDA graph for SqueezeLLM (vllm-project#2161)

  WoosukKwon

  authored
  6f41f0e

• [Minor] Fix a typo in .pt weight support (vllm-project#2160)

  WoosukKwon

  authored
  2c9b638

• [Minor] Add Phi 2 to supported models (vllm-project#2159)

  WoosukKwon

  authored
  f8c688d

• [Minor] Fix xformers version (vllm-project#2158)

  WoosukKwon

  authored
  c9fadda

• [Minor] Add more detailed explanation on quantization argument (vllm-project#2145)

  WoosukKwon

  authored
  30fb095

• Temporarily enforce eager mode for GPTQ models (vllm-project#2154)

  WoosukKwon

  authored
  3a765bd

• [Docs] Add CUDA graph support to docs (vllm-project#2148)

  WoosukKwon

  authored
  26c52a5

• Remove dependency on CuPy (vllm-project#2152)

  WoosukKwon

  authored
  c3372e8

• Pin PyTorch & xformers versions (vllm-project#2155)

  WoosukKwon

  authored
  b0a1d66

• Fix all-reduce memory usage (vllm-project#2151)

  WoosukKwon

  authored
  e1d5402

• [Minor] Delete Llama tokenizer warnings (vllm-project#2146)

  WoosukKwon

  authored
  3d1cfbf

• Optimize model execution with CUDA graph (vllm-project#1926)

  Show description for 37ca558
  

  3 peopleauthored
  37ca558

Commits on Dec 16, 2023

• [ROCm] Temporarily remove GPTQ ROCm support (vllm-project#2138)

  WoosukKwon

  authored
  2acd76f

Commits on Dec 15, 2023

• [Docs] Add supported quantization methods to docs (vllm-project#2135)

  WoosukKwon

  authored
  b81a6a6

Commits on Dec 14, 2023

• Bump up to v0.2.5 (vllm-project#2095)

  WoosukKwon

  authored
  31c1f32

Commits on Dec 13, 2023

• [BugFix] Fix input positions for long context with sliding window (vllm-project#2088)

  WoosukKwon

  authored
  f1c8520

• [Docs] Add notes on ROCm-supported models (vllm-project#2087)

  WoosukKwon

  authored
  096827c

• Update installation instruction for vLLM + CUDA 11.8 (vllm-project#2086)

  WoosukKwon

  authored
  6565d9e

• Implement lazy model loader (vllm-project#2044)

  WoosukKwon

  authored
  518369d

• Fix peak memory profiling (vllm-project#2031)

  WoosukKwon

  authored
  30bad5c

Pagination

PreviousNext