vllm-project/vllm

Pull Requests Commits

jeejeelee committed 19 days ago

931b3f61

[ROCm][Engine] Fix GPU memory leaks in engine shutdown and test workaround for async KV prefix cache reset (#38503)

AndreasKaratzas committed 19 days ago

Verified 95995bbe

[Feature] Warm up readonly multimodal processor during renderer startup (#40797)

fake0fan committed 19 days ago

Verified 07351e08

[ROCm][CI] Fix `trust_remote_code` AttributeError in EAGLE3 acceptance length test (#40306)

AndreasKaratzas committed 19 days ago

Verified 428b988c

[ROCm][CI] Fix TestSiluMulGroupFp8QuantModel after W8A8 block linear refactor (#39799)

AndreasKaratzas committed 19 days ago

Verified e54894fc

[Test] Increase qwen2_vl num_logprobs to fix torch 2.12 update (#40818)

angelayi committed 19 days ago

Verified bc2ae5a3

[Refactor] Remove unused dead code (#40640)

yewentao256 committed 20 days ago

Verified a474da28

[BE][Bugfix] Respect TORCH_COMPILE_DISABLE env var at the vLLM config level for torch 2.12 (#40715)

Lucaskabela committed 20 days ago

Verified ce6a199e

[Bugfix] add seq_lens_cpu_upper_bound to CommonAttentionMetadata in mla_runner.py (#40844)

ignaciosica committed 20 days ago

Verified f88763ef

[EPLB] Fix replica selection bias in fused_moe router (#40810)

arpera committed 20 days ago

Verified 333529de

[Bugfix][CI] Fix wrong residual shape in TestFusedAddRMSNorm.example_inputs that causes flaky test (#40629)

zhangj1an committed 20 days ago

Verified 88256082

[Bug] Fix GLM-5.1 running error on ROCm platform (#40763)

qli88 committed 20 days ago

Verified 095d2f87

[Build] Add Python 3.14 to supported version list. (#34770)

nascheme committed 20 days ago

Verified 21792520

[Frontend] Delegate to vLLM Omni When `--omni` Passed (#40744)

alex-jw-brooks committed 20 days ago

Verified 5e11b403

[Docs] Add docs for context extension using the yarn method (#37430)

labAxiaoming committed 20 days ago

Verified f768b447

[Refactor] Unify 2D/3D kernels in triton_unified_attention (#40631)

JartX committed 20 days ago

Verified 914d0464

[Quantization] add humming quantization kernel (#34556)

jinzhen-lin committed 20 days ago

Verified 9f771b3a

fused_moe: treat NIXL EP as batched experts (#40412)

itayalroy committed 20 days ago

Verified c9d3c6e6

[kv_offload+HMA][9/N]: Support lookup with multiple KV groups (#39401)

orozery committed 20 days ago

Verified 51adca74

[Bugfix][MoE] Unpad routed output before shared expert add [Fixes #35949] (#40794)

netanel-haber committed 20 days ago

Verified e8eb0490

[Attention] use diff kv backend for mimo v2 flash (#40045)

ZJY0516 committed 20 days ago

Verified e8ee2a78

[Bugfix][Parser] Fix Mistral tool parser for HF tokenizers (#39294)

thomasmaindron committed 20 days ago

Verified 2ec18f5d

[Build] Bump CUDA to 13.0.2 to match PyTorch 2.11.0 (#40669)

dmitry-tokarev-nv committed 20 days ago

Verified 6dec49f2

[CI/Build] Add e2e test for ViT CUDA graph (#40780)

shen-shanshan committed 20 days ago

Verified b5587e10

Fix Nano Nemotron VL static image inputs (#40724)

milesial committed 20 days ago

Verified 9ad5abe7

[Bugfix] Fix IMA in DSA + MTP (#40772)

WoosukKwon committed 20 days ago

Verified 7d3195ea

[Model] Gemma4: add bidirectional vision attention for sliding layers with window guard (#40534)

lucianommartins committed 20 days ago

Verified 512f5221

[XPU] Enable torch.compile for XPU GDN attention (#39466)

yuwenzho committed 20 days ago

Verified 4c34b2f6

Support only half types for concat_mla_q kernel (#37892)

xyang16 committed 20 days ago

Verified cf8a613a

[XPU][CI] Fix Docker cleanup races on Intel CI runners (#40761)

zxd1997066 committed 20 days ago

Verified 01acf96c

Older