[Core][NIXL] Support HMA+NixlConnector#31802

Closed

NickLucche wants to merge 75 commits intovllm-project:mainfrom

NickLucche:nixl-hma2

Commits on Dec 14, 2025

Commits on Dec 15, 2025

nits
ivanium
committed

Commits on Dec 17, 2025

[CI] Generalize gsm8k test args and add Qwen3-Next MTP B200 test (#30723)

mgoin
authored and
NickLucche
committed
[Frontend] Add `max-completion-token` option to transcription/translation endpoints (#30769)
NickLucche
committed
[Refactor] Small refactor for group topk (#30562)

authored and
NickLucche
committed
[Perf] Do FP4 quant before All gather on flashinfer trtllmgen MOE (#30014)

jiahanc
authored and
NickLucche
committed
[Attention] Cache attention metadata builds across hybrid KV-cache groups (#29627)

authored and
NickLucche
committed
[Core][MM] Optimize encoder cache manager by operating with embeddings only (#30475)

authored and
NickLucche
committed
[Bugfix][DSV32] Fix overflow in topk. (#30754)

authored and
NickLucche
committed
[Kernel][Quantization][MoE] add marlin kernel support for turing (sm75) (#29901)

authored and
NickLucche
committed
[CI] Skip ci failure test (#30804)

yewentao256
authored and
NickLucche
committed
[Perf][Kernels] Vectorize `csrc/activations_kernels.cu` (#29512)

mgoin
authored and
NickLucche
committed
[ROCm] [Bugfix] Fix torch sdpa hallucination (#30789)

tjtanaa
authored and
NickLucche
committed
Replace deprecated enable_fusion with fuse_norm_quant in test_rms_group_quant (#30817)

mgoin
authored and
NickLucche
committed
[MM] Pass FA version in ViT Attn (#30756)

NickLucche
and
DarkLight1337
committed
[docker] Allow kv_connectors install to fail on arm64 (#30806)

Amr Mahdi
authored and
NickLucche
committed
Fix nemotron_nas intermediate_size computation (#30795)

grzegorz-k-karch
authored and
NickLucche
committed
Update model-hosting-container-standards to 0.1.10 (#30815)

mgoin
authored and
NickLucche
committed
[CPU] Add action to automatically label CPU related PRs (#30678)

fadara01
authored and
NickLucche
committed
[CI/Build] Fix compatibility between #30244 and #30396 (#30787)

DarkLight1337
authored and
NickLucche
committed
bump up compressed tensors version to 0.13.0 (#30799)

authored and
NickLucche
committed
Update note comment for flashinfer attention warmup (#30711)

mgoin
authored and
NickLucche
committed
[Bugfix][CPU] Fix CPU backend ROPE dispatch for VL models (#30829)

authored and
NickLucche
committed
[XPU] fix broken fp8 online quantization for XPU platform (#30831)

yma11
authored and
NickLucche
committed
[Bugfix][Frontend] Prevent IndexError in MiniMax M2 tool parser during streaming extraction (#30555)

WangErXiao
authored and
NickLucche
committed
[Mamba] Removed disable cascade attn in MambaModelConfig (#30712)

Josephasafg
authored and
NickLucche
committed
CustomOp: grouped topk (#29575)

xinyu-intel
authored and
NickLucche
committed
[NIXL][Bugfix] Fix NIXL/RDMA registration failure over CuMemAllocator (#29569)

Somoku
and
NickLucche
committed
[Doc][ResponsesAPI] add documentation (#30840)

authored and
NickLucche
committed
[Kernels][FI] Skip trtllm attention when num_kv_heads=1 (#30842)

yeqcharlotte
authored and
NickLucche
committed
[UX] Make `vllm bench serve` discover model by default and use --input-len (#30816)

mgoin
authored and
NickLucche
committed
[compile] Disable aot when eager backend is used. (#30810)

zhxchen17
authored and
NickLucche
committed
[compile] Ignore VLLM_FORCE_AOT_LOAD from cache factors (#30809)

zhxchen17
authored and
NickLucche
committed
[Fix]Load kv-cache dtype from hf_quant_config.json automatically (fix for reverted PR) (#30785)

authored and
NickLucche
committed
[compile] Recompile graph module during Dynamo cache loading. (#30743)

zhxchen17
authored and
NickLucche
committed
[Bug] Fix AttributeError: 'ColumnParallelLinear' object has no attribute `weight_scale_inv` (#30823)

yewentao256
authored and
NickLucche
committed
[Refactor] [4/N] Move VLLM_SERVER_DEV endpoints into the serve directory (#30749)

chaunceyjiang
authored and
NickLucche
committed
[ci] Sync test areas yaml file with test-pipeline (#30862)

khluu
authored and
NickLucche
committed
[Bugfix] deepseek-V3.2 self.weights_proj has no bias (#30841)

baoqian426
authored and
NickLucche
committed
Fix lazy import (#30858)

hmellor
authored and
NickLucche
committed
chores: adjust the attn register param order (#30688)

ILikeIneine
authored and
NickLucche
committed
[Fix] uniform decode batch check (#30747)

Jialin
authored and
NickLucche
committed
[Docs] fix function name (#30748)

lengrongfu
authored and
NickLucche
committed
Adapt the old parameter enable_thinking in chat_template_kwargs (#30852)

authored and
NickLucche
committed
[Model] Gemma3: Support untied word embeddings (#30827)

www-spam
authored and
NickLucche
committed
wip
NickLucche
committed
wip
NickLucche
committed

Commits on Dec 18, 2025

wip
NickLucche
committed

Commits on Dec 22, 2025

Commits on Jan 6, 2026