[Bugfix] Count per-group blocks in get_max_concurrency_for_kv_cache_config (#48317)

ormandj committed 3 months ago

[Cleanup] Remove unused StructuredOutputRequest.status field (#49235)

Nick Hill committed 3 months ago

[Attention][MLA][DCP] Query replication for MLA decode (DeepSeek-V2/R1 + Kimi-K2.5) (#45964)

sungsoo ha committed 3 months ago

[RL Infra][FlashInfer] Enable router replay output from FlashInfer monolithic MoE kernel (#44214)

Xuanyu Zhang committed 3 months ago

[copy of #45208] CuMem slept-L1 fragmentation accounting (#49208)

Matthew Bonanni committed 3 months ago

[chore] adjust logo be more friendly to white background terminal (#48938)

Ning Xie committed 3 months ago

[ROCm] [Release] [Bugfix] Fix the per commit wheel release pipeline. (#49245)

TJian committed 3 months ago

Support loading sample_from_anchor flag from speculators config (#48639)

Fynn Schmitt-Ulms committed 3 months ago

[ROCm][CI] fix test_rocm_quick_reduce.py (#49234)

Charlie Fu committed 3 months ago

[Bugfix][Spec-Decode] Populate draft seq_lens_cpu_upper_bound for spec-decode attention metadata (#44492)

Oxana Korzh committed 3 months ago

[Feat][Perf] Add new warmup infrastructure for JITs (#47451)

Roberto L. Castro committed 3 months ago

[Bugfix][KV Offload] Preserve reachable tails for hybrid SWA groups (#48911)

coltonottley committed 3 months ago

Cosmos3 FP8 ModelOpt/Diffusers remapping (#48952)

wkutak committed 3 months ago

[Revert][Bugfix] Restore MiniCPM-V 4.6 ViT QKV weight loader (#49193)

tc-mb committed 3 months ago

[Attention] FlashAttention 4 SM100 FP8 kv cache support (#42569)

Matthew Bonanni committed 3 months ago

[XPU]add sycl path for Mhc (#47245)

Liangqiusong committed 3 months ago

[Rust Frontend] Bump `xgrammar-structural-tag` and enable local extension (#49161)

Bugen Zhao committed 3 months ago

[Multimodal] Allow keeping original image mode for ImageIO (#49159)

Isotr0py committed 3 months ago

[KV Offload] Support self-describing KV events with TieringOffloadingSpec (#48679)

Chang Guo committed 3 months ago

[Bugfix][Gemma4] Fix ModelOpt mixed-precision MoE config mapping (#48563)

wangqian committed 3 months ago