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

Chenheli Hua committed about 1 year ago

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

Lucas Wilkinson committed about 1 year ago

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

Michael Goin committed about 1 year ago

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

amirkl94 committed about 1 year ago

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

Michael Goin committed about 1 year ago

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

Woosuk Kwon committed about 1 year ago

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

Marko Rosenmueller committed about 1 year ago

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

bnellnm committed about 1 year ago

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

Woosuk Kwon committed about 1 year ago

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

22quinn committed about 1 year ago

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

Ruixiang Tan committed about 1 year ago

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

Isotr0py committed about 1 year ago

Migrate LlavaOnevisionMultiInputs to TensorSchema (#21844)

Benji Beck committed about 1 year ago

Add return_token_ids parameter to OpenAI API endpoints (#22587)

Yuge Zhang committed about 1 year ago

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

yiz-liu committed about 1 year ago

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

TJian committed about 1 year ago

[Bugfix] Fix benchmark_moe.py (#23177)

Jee Jee Li committed about 1 year ago

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

myselvess committed about 1 year ago

[Model] Add multi_label_classification support (#23173)

wang.yuqi committed about 1 year ago

[NVIDIA] Support Flashinfer TRTLLM FP8-q/kv/out Attention Kernel (#21716)

elvischenv committed about 1 year ago