[Doc] Improve serve parameter documentation with meaningful defaults (#33082)

Karan Bansal committed 10 months ago

Support compress-tensors with nvfp4 or fp8 weights and modelopt with nvfp4 weights on Turing (#33076)

IriKa committed 10 months ago

[BugFix] Fix P/D with non-MoE DP (#33037)

Nick Hill committed 10 months ago

Support heterogeneous NemotronHPuzzle model (#32549)

danielafrimi committed 10 months ago

[LoRA][Spec Decode] Support LoRA for Nemotron-H MTP models (#32265)

danisereb committed 10 months ago

[Frontend] Cleanup api server (#33158)

wang.yuqi committed 10 months ago

[Metrics][MFU] Fix UnembedMetrics FLOP overcounting for prefill (#33045) (#33045)

omkhalil committed 10 months ago

[Bugfix] Fix DeepseekV32 `AssertionError: num_kv_heads == 1` (#33090)

Nicolò Lucchesi committed 10 months ago

[5/N][Attention] Finish eliminating `vllm/attention` folder (#32064)

Matthew Bonanni committed 10 months ago

[Bugfix] Disable CG for Whisper+FA2 (#33164)

Nicolò Lucchesi committed 10 months ago

[Metrics] [KVConnector] Add Offloading Connector metrics (#27942)

omerpaz95 committed 10 months ago

Fix weight mapping test for Transfomers v5 (#33162)

Harry Mellor committed 10 months ago

[Frontend] Frontend will only attach supported tasks corresponding entrypoints. (#33139)

wang.yuqi committed 10 months ago

[AMD][QWEN3-NEXT] FP8 Tunings (#32042)

Lifan Shen committed 10 months ago

[AMD][Kernel][BugFix] Use correct scale in concat_and_cache_ds_mla_kernel when on gfx942 (#32976)

rasmith committed 10 months ago

[Models] Kimi-K2.5 (#33131)

Roger Wang committed 10 months ago

[CI][Pooling] Stabilize ModernBERT test (#32909)

Andreas Karatzas committed 10 months ago

[code clean] remove duplicate code (#33135)

Ning Xie committed 10 months ago

[Frontend] Cleanup serving engine (#33103)

Cyrus Leung committed 10 months ago

[torch.compile] Stop assuming 32 bit indexing (#33113)

Richard Zou committed 10 months ago