[CI] Add ci-fetch-log.sh helper for Buildkite job logs (#41517)

Michael Goin committed 6 months ago

Revert "[Build] Make bundled DeepGEMM wheel portable across Python versions" (#41512)

Michael Goin committed 6 months ago

[Build] Switch CUDA 13.0 wheel builds to PyTorch manylinux_2_28 base (#41416)

Michael Goin committed 6 months ago

[Bugfix][Gemma 4] Clamp soft-token estimate to max_soft_tokens (#40796)

Hoang Nguyen committed 6 months ago

Refactor Step3Text loading to use AutoWeightsLoader (#41492)

Matthew Santiago committed 6 months ago

[Doc] Add Codex usage example (#41358)

Chauncey committed 6 months ago

[ROCm][Bugfix] Fix init-time bias dtype cast when gate.out_dtype is None (#41405)

Rita Brugarolas committed 6 months ago

[vLLM IR] 2/N fused_add_rms_norm and maybe_inplace overload (#36823)

Luka Govedič committed 6 months ago

Limit concurrency on `test_transcription_api_correctness.py` (#41478)

Ekagra Ranjan committed 6 months ago

[MM][CG] Support ViT CG for Qwen2.5-VL (#40830)

John Calderon committed 6 months ago

[Fix] Sync gemma4 chat template from hf (#39570)

FredericOdermatt committed 6 months ago

Re-enable allreduce rms fusion for DP / PP (#41458)

Andy Lo committed 6 months ago

[Build] Make bundled DeepGEMM wheel portable across Python versions (#41476)

Michael Goin committed 6 months ago

[Bugfix] Fix persistent_topk inter-CTA init race on RadixRowState (#41444)

Yongye Zhu committed 6 months ago

[DSV4] Add knob to enable pre-attn gemm (#41443)

Yongye Zhu committed 6 months ago

[Perf] Intergrate Tile Kernels `head_compute_mix_kernel` for Deepseek-V4 (#41255)

Isotr0py committed 6 months ago

[Attention] Abstract the MLA prefill backends and eliminate cuDNN (#32623)

Matthew Bonanni committed 6 months ago

[Model Runner V2] Add `logprob_token_ids` support (#40559)

Wentao Ye committed 6 months ago

[Eval][CI] Add basic mrcr eval to tests/evals/ (#40164)

Michael Goin committed 6 months ago

[ROCm][FEAT] AITER Fused Allreduce + RMSNorm (#37646)

vllmellm committed 6 months ago