vllm-project/vllm

Pull Requests Commits

Use uv to install python in Dockerfile

mgoin committed 88 days ago

728c365e

Change size of single CUDA graph for CI to 4 (#26089)

tdoublep committed 88 days ago

Verified be8921fb

Update base image to 22.04 (jammy) (#26065)

huydhn committed 88 days ago

Verified d4e7a115

Run:ai model streamer add GCS package support (#24909)

pwschuurman committed 88 days ago

Verified be22bb6f

[Misc] Make handling of SamplingParams clearer in n>1 case (#26032)

njhill committed 88 days ago

Verified 169313b9

[ROCm][Bugfix] Add missing parameter to ROCm backend (#26029)

gshtras committed 88 days ago

Verified 0b018d8b

Support RL online quantization with torchao (#23014)

jerryzh168 committed 88 days ago

Verified c3124680

[BugFix] ChunkedLocalAttention is currently not CG compatible (#26034)

LucasWilkinson committed 88 days ago

Verified 4134312b

[Bug] Fix Negative Cuda Memory Usage (#25683)

yewentao256 committed 88 days ago

Verified da554f93

[ROCm][Build] Add support for AMD Ryzen AI MAX / AI 300 Series (#25908)

hyoon1 committed 88 days ago

Verified aac622e0

[BugFix][DP/EP] Fix CUTLASS MLA hang under load (#26026)

LucasWilkinson committed 88 days ago

Verified 1726e93e

[CI] Tweaks to GPT-OSS Eval (Blackwell) for stability (#26030)

mgoin committed 88 days ago

Verified ee04c0cd

Fix test_mamba_ssm_ssd.py due to missing _query_start_loc_to_chunk_indices_offsets (#25995)

hl475 committed 88 days ago

Verified c36f0aa3

[NVIDIA] Blackwell Family (#24673)

johnnynunez committed 88 days ago

Verified 5234dc74

[Bugfix] Apply same sampling parameters for both `n=1` and `n>1` (#26005)

kmaehashi committed 89 days ago

Verified 3b7c20a6

[Benchmark] Finish documented v0.11.0 deprecation of --endpoint-type (#26007)

natoscott committed 89 days ago

Verified f9e71481

[MISC] Fix misleading batch_size_capture_list when cuda_graph_sizes < 4 (#25829)

billishyahao committed 89 days ago

Verified 2518230d

[CI] Only capture a single CUDA graph size in CI by default (#25951)

hmellor committed 89 days ago

Verified a332b845

[Misc] Factor out common `_apply_feature_select_strategy` (#26003)

DarkLight1337 committed 89 days ago

Verified 1405f0c7

[BugFix][MM] Fix Nonetype error when video is cache in qwen2.5-omni-thinker (#26004)

wwl2755 committed 89 days ago

Verified 84d57342

[Doc] updating torch.compile doc link (#25989)

vnadathur committed 89 days ago

Verified 57b46d76

[Misc]allow disable pynccl (#25421)

luccafong committed 89 days ago

Verified f48b6a03

Update to Transformers `v4.56.2` (#24638)

hmellor committed 89 days ago

Verified 2a69ab48

[BugFix] Fix default kv-cache-dtype default for DeepseekV3.2 (#25988)

LucasWilkinson committed 89 days ago

Verified 8d7da92f

[Bugfix] Fix `__syncwarp` on ROCM (#25996)

zhewenl committed 89 days ago

Verified e952eee6

[MM] Add text-only mode for Qwen3-VL (#26000)

ywang96 committed 89 days ago

Verified 66bca9b8

Fix INT8 quantization error on Blackwell GPUs (SM100+) (#25935)

certainly-param committed 89 days ago

Verified 99028fda

[Log] Optimize Log for FP8MOE (#25709)

yewentao256 committed 89 days ago

Verified 12449488

Update launch_bounds_utils.h for correct compile on Multiple Cuda Arch - PTXAS out of range Warning (#25843)

DrStone1971 committed 89 days ago

Verified a73f6491

[Model] MTP fallback to eager for DeepSeek v32 (#25982)

luccafong committed 89 days ago

Verified 001e50c9

Older