fix(openai): tolerate empty content in forced tool choice (#40148)

jack committed 6 months ago

nixl refactor: new transfer design (#40731)

zhanqiuhu committed 6 months ago

[MM][Gemma4] Respect max_soft_tokens in encoder budget (#41799)

SeongJun Lee committed 6 months ago

[CPU][RISC-V] Auto-bind OMP threads and harden nobind path (#40569)

lyd1992 committed 6 months ago

Fix some legacy checkpoints with deprecated `rope_type` values (#41734)

Harry Mellor committed 6 months ago

[Bugfix][Rocm]Aiter MoE re-uses existing tensor addresses after weight update. (#40390)

Yuankai Chen committed 6 months ago

[Bugfix][CI] Fix Disaggregated test area path (#41794)

Nicolò Lucchesi committed 6 months ago

fix(rocm): remove workaround causing invalid argument on Qwen3.5 with TP=2 (#40686)

Zhaodong Bing committed 6 months ago

[XPU] Disable CUDA graph memory estimate on XPU platform (#41344)

Chaojun Zhang committed 6 months ago

[Examples] Resettle Disaggregated examples. (#40759)

wang.yuqi committed 6 months ago

[Bugfix] Fix codegen for unqualified names (#40726)

Lucas Kabela committed 6 months ago

[Feat] dnnl build for AVX2 W8A8 Int8 (#41318)

Tianmu Li committed 6 months ago

[CPU] Add FP8 W8A16 linear support (#41186)

Yuwen Zhou committed 6 months ago

[Bugfix] Skip PP sampled-token receive on last rank during async scheduling (#40749)

wi-adam committed 6 months ago

[Spec Decode] Allow multimodal models with a warning (#41752)

Li committed 6 months ago

[ROCm][CI] Refine gating tests (#37243)

Andreas Karatzas committed 6 months ago

[Fix] Add missing stubs from cpu fp8 attention changes (#41387)

Tianmu Li committed 6 months ago

[Build] Fall back to system libgomp when torch has no vendored copy (#40575)

lyd1992 committed 6 months ago

[Mistral Tokenizer] allow more leniency in apply_chat_template (#41658)

Julien Denize committed 6 months ago

[Frontend] Supports resubmitting output items with missing fields in Responses API (#41355)

Chauncey committed 6 months ago