[XPU] [BugFix] Add deepseek_v4_fp8 to xpu supported_quantization list (#50434)

Xiaochang Wu committed 3 months ago

[XPU] Fix FP8 block scale layout for MLA compatibility (#50349)

Ma Jian committed 3 months ago

[CI/Build][AMD] Install triton_kernels via CMake (#50328)

Ryan Rock committed 3 months ago

[ROCm] Add tuned selective_state_update float16 config for AMD Instinct MI325X (#50006)

vanshbhatia-amd committed 3 months ago

[ROCm][CI] Use larger atol value for INT3 in test_quick_all_reduce.py (#50450)

music-dino committed 3 months ago

[Hardware][AMD][Kernel][CI][Bugfix] Fix ROCm DeepEP FP8 max (#50467)

Matt committed 3 months ago

[CI] Retry Buildkite API rate limits (#50481)

Kevin H. Luu committed 3 months ago

[Frontend][Bugfix] Use default tool call IDs for Kimi K3 for conversation-level uniqueness (#50420)

Bugen Zhao committed 3 months ago

[Bugfix][ROCm] AITER MLA: size MTP verification decode metadata for real qlen/dtype (#45227)

chaeminlim-mb committed 3 months ago

Enable gfx1250 ROCm architecture (#46516)

Juan Villamizar committed 3 months ago

[CI] Retry Hugging Face processor loading (#49908)

Andreas Karatzas committed 3 months ago

[Bugfix][Rust Frontend] Select earliest-completing stop string (#50200)

Samuel Laferriere committed 3 months ago

[ROCm]: bump AITER to 0.1.19 (#49361)

Rohan Potdar committed 3 months ago

[Kimi K3 Bug] Fix deepgemm support for kimi k3 (#50458)

Wentao Ye committed 3 months ago

[CI] Fix `tests/entrypoints/multimodal/openai/chat_completion/test_audio.py::test_chat_streaming_audio` (#50451)

Nicolò Lucchesi committed 3 months ago

[Model Runner V2][Spec Decode] Add multi-layer MTP speculator (#48892)

Giancarlo Delfin committed 3 months ago

[CI] Retire the v1 PR label rule, add mrv2 (#50475)

jcotant-inferact committed 3 months ago

[Bugfix] Preserve Marlin runtime tensor storage across weight reload (#48438)

Ryan Clark committed 3 months ago

[Rust Frontend] Improve startup failure and readiness logs (#50406)

Bugen Zhao committed 3 months ago

[Kernel][Helion] Disable unsafe B200 RMS reduction warp specialization (#50345)

Shangdi Yu committed 3 months ago