vllm-project/vllm

Pull Requests Commits

[bug] Fix "Current vLLM config is not set." warnings when FlashInfer attention is used (#30241)

nvpohanh committed 18 days ago

Verified eea41804

[Perf] Enable environment cache in EngineCore to enable the feature for UniProcExecutor as well (#29289)

Jialin committed 18 days ago

Verified 9f042ba2

{Deprecation] Remove tokenizer setter (#30400)

DarkLight1337 committed 18 days ago

Verified e72d65b9

[P/D] KV Load Failure Recovery/Abort Configuration (#26813)

wseaton committed 18 days ago

Verified a9e4106f

[Bugfix] Fix HunyuanOCR cross-image contamination in batch processing (#30344)

anker-c2 committed 18 days ago

Verified e8e8cd73

[Chore] Delay recent deprecations (#30398)

DarkLight1337 committed 18 days ago

Verified 253305d5

[Misc] Consistent case for `vllm bench serve` results (#30403)

MatthewBonanni committed 18 days ago

Verified 794a7875

[Docs] Generate full list of metrics in user docs (#30388)

markmc committed 18 days ago

Verified 2dcbac90

[BugFix] Fix `AttributeError: 'MergedColumnParallelLinear' object has no attribute 'weight_scale'` (#30399)

LucasWilkinson committed 18 days ago

Verified aacf0abf

[Core] Whisper enable `FULL_DECODE_ONLY` CudaGraph (#30072)

NickLucche committed 18 days ago

Verified c756fb67

[BugFix] Fix minimax m2 model rotary_dim (#30384)

rogeryoungh committed 18 days ago

Verified d017bceb

[CPU] Support for Whisper (#30062)

aditew01 committed 18 days ago

Verified cebda2a4

[Bugfix] tpu_model_runner: set vllm config context when calling reset_dynamo_cache() (#30331)

dtrifiro committed 18 days ago

Verified 53d2420b

[Bugfix] Fix the issue where DeepSeek v3.2 cannot use structured_output (#30371)

chaunceyjiang committed 18 days ago

Verified 9db78f34

[cpu][ci] Add CPU Attention Tests for Neon Backend (#30347)

fadara01 committed 18 days ago

Verified 434ac76a

[ROCm][CI] Attempt to fix the failures under a subgroup of the e2e the test group (#29358)

AndreasKaratzas committed 18 days ago

Verified ed7af317

[CMake][Build]: Remove unused ACL CMake env variables (#30339)

Radu2k committed 18 days ago

Verified 18034580

[Bugfix] Cache added_vocab to avoid per-token overhead (#30351)

scratch-ml committed 18 days ago

Verified d007387a

Fix typos in comments across multiple files (#30345)

wilsonwu committed 18 days ago

Verified 3bdd4266

[bugfix][quantization] fix quark qwen3 kv_cache quantization (#30308)

haoyangli-amd committed 18 days ago

Verified 06462392

[CI] Reduce Flakiness For test_spec_decode.py::test_suffix_decoding_acceptance (#30367)

micah-wil committed 19 days ago

Verified 7d80c73d

[CI/Build][AMD] Skip quantization kernels tests that require CUTLASS or e4m3fn when not supported by platform (#30020)

rasmith committed 19 days ago

Verified b75f826f

[responsesAPI][6] Fix multi turn MCP tokenization (#30230)

qandrew committed 19 days ago

Verified c3487aca

[Attention] Make seq_lens_cpu optional in CommonAttentionMetadata to enable true async spec-decode (#29624)

LucasWilkinson committed 19 days ago

Verified abe93bce

[Bugfix] Fix fp8 DeepGemm compilation issues (#30336)

ElizaWszola committed 19 days ago

Verified 2e7035dd

[Bugfix] Fix cuda graph sizes when running with speculative decoding (#30330)

PatrykSaffer committed 19 days ago

Verified 4c2e10ea

[V1][Spec Decode] Optimize Medusa proposer to avoid GPU-CPU sync (#29723)

dongbo910220 committed 19 days ago

Verified 03b5f940

Improve wvsplitK tile and balance heristics. (#29937)

amd-hhashemi committed 19 days ago

Verified 2e7054da

[Rocm][torch.compile] Adding layernorm + fp8 block quant and silu + fp8 block quant for Aiter (#25693)

charlifu committed 19 days ago

Verified 3c680f4a

[Quantization] FP8 Weight Reloading for Quantized RL Rollout (#28480)

kylesayrs committed 19 days ago

Verified fccd5325

Older