Fix Mistral yarn warning in Transformers v5 (#37292)

Harry Mellor committed 7 months ago

Automatically add links to API docs for matching strings in docs (#37434)

Harry Mellor committed 7 months ago

[KV Offload] Clean up ARC/LRU refactoring leftovers: group ARC tests and fix stale comment (#38217)

Ronen Schaffer committed 7 months ago

[Bug] Fix Trtllm Fp8 MoE Weight Shuffle Memory Fragamentation (#39054)

Wei Zhao committed 7 months ago

[Bugfix] Correct mistake in chained comparison in static assert logic (#38699)

Kyle Mylonakis committed 7 months ago

only patch runtime_env for torch >= 2.10 (#38763)

Rohan Potdar committed 7 months ago

[vLLM IR] rework gemma_rms_norm (#39014)

Jiangyun Zhu committed 7 months ago

[Model] Use AutoWeightsLoader for FalconH1 (#39092)

Rishapveer Singh committed 7 months ago

nano-nemotron-vl: get_mm_max_tokens_per_item for audio, video, image == seq_len (#38727)

Netanel Haber committed 7 months ago

[ROCm] Remove unused IS_FNUZ parameter from reshape_and_cache_shuffle_kernel (#39123)

Andrew Barnes committed 7 months ago

[ROCm][CI] Fix test repo-root assumptions (#39053)

Andreas Karatzas committed 7 months ago

[Kernels][MoE] Fix legacy_routing to use bitmatrix-based routing path (#38504)

Andreas Karatzas committed 7 months ago

[BugFix][MRV2] Fix cuda event reuse race (#39115)

Nick Hill committed 7 months ago

[MoE Refactor] Split up compressed_tensors_moe.py (#38960)

bnellnm committed 7 months ago

[NVFP4] Support NVFP4 dense models from `modelopt` and `compressed-tensors` on AMD Instinct MI300, MI355X and Hopper through emulation (#35733)

fxmarty-amd committed 7 months ago

[Attention][MLA] Re-enable FA4 as default MLA prefill backend (#38819)

Matthew Bonanni committed 7 months ago

[MRV2] Fix hanging issue with DeepSeek V3.2 by setting `skip_attn=False` (#39098)

Woosuk Kwon committed 7 months ago

NemotronH default mamba_ssm_cache_dtype=float32; enable auto-hook for NemotronHNanoVLV2Config (#39032)

Netanel Haber committed 7 months ago

[Quantization] Add FlashInfer CuteDSL batched experts backend for NVFP4 MoE (#38251)

Yongye Zhu committed 7 months ago

[EASY] Drop duplicate KV-cache initialization (#38799)

namgyu-youn committed 7 months ago