vllm-project/vllm

Pull Requests Commits

fix cuda graph (#22721)

fsx950223 committed 110 days ago

Verified d983769c

[BugFix] Fix stuck stats/metrics after requests are aborted (#22995)

njhill committed 110 days ago

Verified 8fd92092

[CI/Build] Sync multimodal tests (#23181)

DarkLight1337 committed 110 days ago

Verified de7b67a0

[CI/Build] Also check DP in benchmarks throughput script (#23038)

zhewenl committed 110 days ago

Verified f7290232

chore: support pytorch format in lora (#22790)

KilJaeeun committed 110 days ago

Verified 1a3079a1

Fix a performance comparison issue in Benchmark Suite (#23047)

louie-tsai committed 110 days ago

Verified 941f5685

[Attention] Optimize make_local_attention_virtual_batches for Flash Attention (#23185)

linzebing committed 110 days ago

Verified a634733f

[Doc] Update V1 status of various pooling models (#23189)

DarkLight1337 committed 110 days ago

Verified 64ab3c72

[Core] Add torch profiler CPU traces for AsyncLLM. (#21794)

huachenheli committed 110 days ago

Verified e58c5a97

[CI Perf] Only test bfloat16 for tests/compile/test_fusion_all_reduce.py (#23132)

mgoin committed 110 days ago

Verified d46d417b

[Core] Optimize scheduler request removal for single completions (#21917)

chi2liu committed 110 days ago

Verified 0167efe2

[Quantization] Bump Compressed Tensors Version (#23202)

kylesayrs committed 110 days ago

Verified c32e6ad1

[Benchmarks] Add video inputs to ShareGPTDataset. (#23199)

huachenheli committed 110 days ago

Verified 1630cc8d

[BugFix] fix CUTLASS MLA full cudagraph (#23200)

LucasWilkinson committed 110 days ago

Verified 14e2b073

[CI/Build] Replace lm-eval gsm8k tests with faster implementation (#23002)

mgoin committed 111 days ago

Verified 0f4f0191

[NVIDIA] Add SM100 Flashinfer Cutlass MoE fp8 backend (#22357)

amirkl94 committed 111 days ago

Verified a38b8af4

[CI/Build] Add support for Python 3.13 (#13164)

mgoin committed 111 days ago

Verified 21dce80e

[Misc] Minor refactoring for FlashInfer backend (#23147)

WoosukKwon committed 111 days ago

Verified e61bac87

fix: use cache_salt for gpt-oss (#23186)

dr75 committed 111 days ago

Verified 80141bbf

[Bugfix] Fix accuracy issue when using flashinfer cutlass moe, TP=1 and modelopt. (#23125)

bnellnm committed 111 days ago

Verified b94faf9d

[Misc] Enable yapf for FlashInfer backend (#23193)

WoosukKwon committed 111 days ago

Verified 5b5f350d

[Frontend] Add `/collective_rpc` API endpoint (#23075)

22quinn committed 111 days ago

Verified f7cf5b51

[Misc] Fix the benchmark's README and improve the error messages for the benchmark's argument checks (#22654)

tanruixiang committed 111 days ago

Verified 03d4235f

[CI/Build] Update transformers to v4.55.2 (#23093)

Isotr0py committed 111 days ago

Verified d6a1a209

Migrate LlavaOnevisionMultiInputs to TensorSchema (#21844)

bbeckca committed 111 days ago

Verified a70d0bd0

Add return_token_ids parameter to OpenAI API endpoints (#22587)

ultmaster committed 111 days ago

Verified 24f4d1a2

[Model] Removes redundant all-reduce operation in Qwen3MoeSparseMoeBlock (#23169)

yiz-liu committed 111 days ago

Verified 4f510bc2

[FEAT] [Performance] Enable DP for ViT in Qwen2.5VL (#22742)

tjtanaa committed 111 days ago

Verified 1298c677

[Bugfix] Fix benchmark_moe.py (#23177)

jeejeelee committed 111 days ago

Verified 4d9c6199

[Model] support new model ovis2.5 (#23084)

myselvess committed 111 days ago

Verified b87cb97a

Older