Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 30, 2024

• Allow Vllm to run with spec dec where the draft model is sharded over multiple GPUs

  tbarton16

  committed
  86ce463

  Copy full SHA for 86ce463

• [Misc] Adjust max_position_embeddings for LoRA compatibility (vllm-project#8957)

  jeejeelee

  authored
  1cabfce

  Copy full SHA for 1cabfce

• [Core] Make scheduling policy settable via EngineArgs (vllm-project#8956)

  schoennenbeck

  authored
  be76e5a

  Copy full SHA for be76e5a

• [Model] Expose InternVL2 max_dynamic_patch as a mm_processor_kwarg (vllm-project#8946)

  Isotr0py

  authored
  2ae25f7

  Copy full SHA for 2ae25f7

• [Model][LoRA]LoRA support added for MiniCPMV2.6 (vllm-project#8943)

  Show description for 8e60afa
  

  jeejeelee

  and

  DarkLight1337

  authored
  8e60afa

  Copy full SHA for 8e60afa

• [Misc][CI/Build] Include cv2 via mistral_common[opencv] (vllm-project#8951)

  ywang96

  authored
  b6d7392

  Copy full SHA for b6d7392

• [Model] support input embeddings for qwen2vl (vllm-project#8856)

  whyiug

  authored
  e01ab59

  Copy full SHA for e01ab59

Commits on Sep 29, 2024

• [Kernel][Model] Varlen prefill + Prefill chunking support for mamba kernels and Jamba model (vllm-project#8533)

  mzusman

  authored
  f13a07b

  Copy full SHA for f13a07b

• [Frontend] Added support for HF's new continue_final_message parameter (vllm-project#8942)

  danieljannai21

  authored
  6c9ba48

  Copy full SHA for 6c9ba48

• [Misc] Fix typo in BlockSpaceManagerV1 (vllm-project#8944)

  juncheoll

  authored
  1fb9c1b

  Copy full SHA for 1fb9c1b

• [BugFix] Fix seeded random sampling with encoder-decoder models (vllm-project#8870)

  Show description for 31f46a0
  

  njhill

  and

  ywang96

  authored
  31f46a0

  Copy full SHA for 31f46a0

• [Model][LoRA]LoRA support added for MiniCPMV2.5 (vllm-project#7199)

  jeejeelee

  authored
  3d49776

  Copy full SHA for 3d49776

• [Model] Support Qwen2.5-Math-RM-72B (vllm-project#8896)

  zhuzilin

  authored
  bc2ef1f

  Copy full SHA for bc2ef1f

• [Build/CI] Set FETCHCONTENT_BASE_DIR to one location for better caching (vllm-project#8930)

  tlrmchlsmth

  authored
  2e7fe7e

  Copy full SHA for 2e7fe7e

• [CI/Build] Add test decorator for minimum GPU memory (vllm-project#8925)

  DarkLight1337

  authored
  26a68d5

  Copy full SHA for 26a68d5

• [Bugfix] Fix Marlin MoE act order when is_k_full == False (vllm-project#8741)

  Show description for d081da0
  

  ElizaWszola

  and

  tlrmchlsmth

  authored
  d081da0

  Copy full SHA for d081da0

• [Bugfix] Block manager v2 with preemption and lookahead slots (vllm-project#8824)

  sroy745

  authored
  5bf8789

  Copy full SHA for 5bf8789

• [Core] Improve choice of Python multiprocessing method (vllm-project#8823)

  Show description for d153703
  

  russellb

  and

  youkaichao

  authored
  d153703

  Copy full SHA for d153703

• [doc] organize installation doc and expose per-commit docker (vllm-project#8931)

  youkaichao

  authored
  cc27644

  Copy full SHA for cc27644

Commits on Sep 28, 2024

• [Bugfix] Support testing prefill throughput with benchmark_serving.py --hf-output-len 1 (vllm-project#8891)

  heheda12345

  authored
  e585b58

  Copy full SHA for e585b58

• [Frontend] Make beam search emulator temperature modifiable (vllm-project#8928)

  Show description for 090e945
  

  nFunctor

  and

  Eduard Balzin

  authored
  090e945

  Copy full SHA for 090e945

• [CI/Build] Update models tests & examples (vllm-project#8874)

  Show description for e1a3f5e
  

  DarkLight1337

  and

  ywang96

  authored
  e1a3f5e

  Copy full SHA for e1a3f5e

• [Bugfix] Fix PP for Multi-Step (vllm-project#8887)

  varun-sundar-rabindranath

  authored
  19d02ff

  Copy full SHA for 19d02ff

• [Bugfix] Fix code for downloading models from modelscope (vllm-project#8443)

  tastelikefeet

  authored
  39d3f8d

  Copy full SHA for 39d3f8d

• [Misc] Remove vLLM patch of BaichuanTokenizer (vllm-project#8921)

  DarkLight1337

  authored
  b0298aa

  Copy full SHA for b0298aa

• [Bugfix][Intel] Fix XPU Dockerfile Build (vllm-project#7824)

  Show description for 260024a
  

  tylertitsworth

  and

  youkaichao

  authored
  260024a

  Copy full SHA for 260024a

• [misc] fix wheel name (vllm-project#8919)

  youkaichao

  authored
  d86f6b2

  Copy full SHA for d86f6b2

Commits on Sep 27, 2024

• [Core] Priority-based scheduling in async engine (vllm-project#8850)

  schoennenbeck

  authored
  bd429f2

  Copy full SHA for bd429f2

• [misc][distributed] add VLLM_SKIP_P2P_CHECK flag (vllm-project#8911)

  youkaichao

  authored
  18e60d7

  Copy full SHA for 18e60d7

• [Core] Multi-Step + Single Step Prefills via Chunked Prefill code path (vllm-project#8378)

  Show description for c2ec430
  

  varun-sundar-rabindranath

  and

  Varun Sundar Rabindranath

  authored
  c2ec430

  Copy full SHA for c2ec430

• [Bugfix] fix for deepseek w4a16 (vllm-project#8906)

  Show description for c5d5535
  

  LucasWilkinson

  and

  mgoin

  authored
  c5d5535

  Copy full SHA for c5d5535

• [Kernel] AQ AZP 4/4: Integrate asymmetric quantization to linear method (vllm-project#7271)

  ProExpertProg

  authored
  172d1cd

  Copy full SHA for 172d1cd

• [torch.compile] use empty tensor instead of None for profiling (vllm-project#8875)

  youkaichao

  authored
  a9b15c6

  Copy full SHA for a9b15c6

• [TPU] Update pallas.py to support trillium (vllm-project#8871)

  bvrockwell

  authored
  8df2dc3

  Copy full SHA for 8df2dc3

• [Bugfix][VLM] Fix Fuyu batching inference with max_num_seqs>1 (vllm-project#8892)

  Isotr0py

  authored
  6d792d2

  Copy full SHA for 6d792d2

Pagination

Previous

Next