 Ziang Li
|
4cfebbb95f
|
[FlashInfer v0.6.12] Support FlashInfer 4over6 NVFP4 (#25239)
|
2026-06-04 14:35:07 -07:00 |
|
|
|
5bf90ad988
|
Enable DeepGEMM PDL on by default (#23979)
Co-authored-by: b8zhong <b8zhong@users.noreply.github.com>
Co-authored-by: Brayden Zhong <brayden@radixark.ai>
|
2026-06-04 14:13:45 -07:00 |
|
|
Liangsheng Yin
|
cd98d97037
|
Use level-1 (quiet) busy memory check in chunked-prefill and streaming tests (#27303)
|
2026-06-04 17:04:27 -04:00 |
|
|
Liangsheng Yin
|
687cfe9198
|
Enable runtime busy memory check for speculation topk>1 (#27228)
|
2026-06-04 16:46:20 -04:00 |
|
|
Ilia Iliev
|
88a9d513e0
|
[Quant] Support asymmetric weight quant in compressed-tensors WNA16 (#25292)
|
2026-06-04 20:15:47 +00:00 |
|
|
YC Yen-Ching Tseng
|
69623f4b11
|
[AMD] Guard aiter greedy_sample OOB token id (fixes VLM MMMU CI) (#27247)
|
2026-06-04 12:53:58 -07:00 |
|
|
shuwenn
|
efe24704b7
|
[HiCache] feat: truncate mamba prefetch length to available host KV size (#26945)
|
2026-06-05 03:01:37 +08:00 |
|
|
Chengze FanandHaiShaw
|
8e836e7dc9
|
Support optional kwargs in AITER fused_moe runner (#26746)
Co-authored-by: HaiShaw <hixiao@gmail.com>
|
2026-06-04 08:25:03 -07:00 |
|
|
Sam Shleifer
|
133254086b
|
Plug mamba_extra_buffer ping-pong slot leaks (#26941)
|
2026-06-04 21:46:36 +08:00 |
|
|
huangtingwei
|
b97a3dbb46
|
[HiSparse PD & PP]Fix HiSparse compatibility with PP decode (#27258)
|
2026-06-04 21:37:42 +08:00 |
|
|
liuxianglong17
|
8933ec8772
|
fix test cases failed on 5/30 in nightly pipeline (#26775)
|
2026-06-04 20:07:20 +08:00 |
|
|
Mick
|
e4a01d5ee2
|
[diffusion] fix: fix realtime webui recording timeline (#27249)
|
2026-06-04 19:58:56 +08:00 |
|
|
huangtingwei
|
a5c7e9d236
|
[HiCache] fix PD L3 cache hit details from decode responses (#27046)
|
2026-06-04 18:01:21 +08:00 |
|
|
Mick
|
9e2ad6054c
|
[diffusion] feat: speed up lossless realtime rgb transport (#27236)
|
2026-06-04 16:59:23 +08:00 |
|
|
Liangsheng Yin
|
ef170c27b6
|
Add quiet mode for busy mem check (level 1: buffer + dump on leak) (#27238)
|
2026-06-04 04:31:57 -04:00 |
|
|
jacky.cheng
|
7aee2ff31b
|
[AMD] Remove BF16-to-FP32 elementwise cast from compressor GEMM on HIP (#26914)
|
2026-06-04 00:58:02 -07:00 |
|
|
YC Yen-Ching Tseng
|
ff93a576e5
|
[AMD] Minimax M25 : FP8 block-scale GEMM dispatch for ROCm 7.0 on gfx950 (#27111)
|
2026-06-04 00:41:42 -07:00 |
|
|
shuwenn
|
1af53f67b2
|
[mem_cache][2/N] refactor: move SWATokenToKVPoolAllocator to allocator/swa.py (#26676)
|
2026-06-04 15:10:37 +08:00 |
|
|
Leon Gao
|
a10bd785be
|
Reduce mamba prefill allocation overhead (#25000)
|
2026-06-04 15:10:16 +08:00 |
|
|
Alex SunandHaiShaw
|
e4191708c9
|
[Qwen3.5][AMD] Fix shared-expert ×ep_size over-count under allreduce-EP (#26845)
Co-authored-by: HaiShaw <hixiao@gmail.com>
|
2026-06-03 23:04:39 -07:00 |
|
|
Heyang Huang
|
858e5a5109
|
[diffusion] chore: disagg server args, launch helpers, and warmup utils (#26119)
|
2026-06-04 13:40:39 +08:00 |
|
|
Zhonghua Deng
|
e541bc3881
|
[EPD] feat: encoder DP mode with per-rank subprocess workers (#26576)
|
2026-06-04 12:37:41 +08:00 |
|
|
Yinghai Lu
|
71c759ebb7
|
[loader] Reduce transient allocations in NVFP4 MoE setup (#26861)
|
2026-06-03 21:13:25 -07:00 |
|
|
Cheng WanandClaude Opus 4.8
|
10ab7c919f
|
[refactor] Retire DecodeInputBuffers / PrefillInputBuffers in favor of CudaGraphBufferRegistry (#27192)
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
2026-06-03 20:52:56 -07:00 |
|
|
Zaili WangandMa Mingfei
|
3b7a258f63
|
[CPU] upgrade dependent torch ver to PT2.12 (#21456)
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com>
|
2026-06-04 11:04:11 +08:00 |
|
|
王鹤男
|
29d23e198f
|
[diffusion] fix: preserve _explicit_fields across dataclasses.replace in DiffGenerator (#25308)
|
2026-06-04 11:02:23 +08:00 |
|
|
Mick
|
11605767e0
|
[diffusion] optimize: skip unused wanvae halo send copies (#27151)
|
2026-06-04 10:23:01 +08:00 |
|
|
Zhanghengand晟海
|
736263f3dc
|
[UnifiedTree]: Support l3 storage for swa and deepseek v4 (#26881)
Co-authored-by: 晟海 <huangtingwei.htw@antgroup.com>
|
2026-06-04 10:17:34 +08:00 |
|
|
Yuzhen ZhouandJiajun Li
|
e03dfa8182
|
[3/N][Sync sglang-miles] TITO Support (#23751)
Co-authored-by: Jiajun Li <48857426+guapisolo@users.noreply.github.com>
|
2026-06-03 21:45:33 -04:00 |
|
|
Jonny Kong
|
084c6a7e2a
|
Refactor simulated acceptance length generation (#26768)
|
2026-06-03 18:31:32 -07:00 |
|
|
|
f6cd1a9822
|
Add num_waiting_uncached_tokens load metric (#27174)
Co-authored-by: cctry <cctry@fb.com>
Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
|
2026-06-03 18:29:49 -07:00 |
|
|
ybyangandLianmin Zheng
|
687baf9471
|
fix(load-snapshot): avoid duplicate zmq bind in multi-tokenizer mode (#27145)
Co-authored-by: Lianmin Zheng <lianminzheng@gmail.com>
|
2026-06-03 18:24:56 -07:00 |
|
|
|
14ed9b448e
|
Add ZMQ IPv6 support, bench_serving sampling params, and reduce routed_dp_rank log noise (#27180)
Co-authored-by: Hanming Lu <69857889+hanming-lu@users.noreply.github.com>
Co-authored-by: Grigory Sizov <grisha.sizov@gmail.com>
|
2026-06-03 17:49:34 -07:00 |
|
|
Yinghai Lu
|
7bb5c96685
|
Trigger scheduler diagnostics on health failure (#26757)
|
2026-06-03 17:19:55 -07:00 |
|
|
Baizhou Zhang
|
d097cd2212
|
[GLM-5] Apply trtllm MHA kernel for GLM-5 on Blackwell (#21332)
|
2026-06-03 17:16:54 -07:00 |
|
|
Mick
|
3790173b3b
|
[diffusion] fix: avoid flashattention forward context lookup (#27153)
|
2026-06-04 08:11:50 +08:00 |
|
|
Clintandclintg6
|
cfb7fb4fad
|
[AMD] Fix TP2 DeepSeek-R1 nhead=64 MLA decode crash and add nightly coverage (#27188)
Co-authored-by: clintg6 <7388379+clintg6@users.noreply.github.com>
|
2026-06-03 16:56:05 -07:00 |
|
|
Cheng WanandClaude Opus 4.8
|
c9ca56da8c
|
Unify full→SWA index translation in init_forward_metadata; drop pool caches (#27091)
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
|
2026-06-03 16:12:27 -07:00 |
|
|
Cheng Wan
|
61aa3293d3
|
Revert "Fix TokenizerManager crash on top_logprobs with tensor values" (#27187)
|
2026-06-03 14:53:28 -07:00 |
|
|
ishandhanani
|
978fb6ed1a
|
hicache kv events: publish split write-through fragments (#27072)
|
2026-06-03 14:52:25 -07:00 |
|
|
jacky.cheng
|
1dd9432889
|
[AMD] Fuse compress norm+rope+hadamard into single Triton kernel (#26894)
|
2026-06-03 14:20:40 -07:00 |
|
|
|
c670609ac5
|
[NPU] Diffusion CI Ground Truth Generation (NPU) (#24630)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Elizaveta Martirosian <you@example.com>
Co-authored-by: Elizaveta Martirosian <elizaveta.martirosian@gmail.com>
Co-authored-by: ronnie_zheng <zl19940307@163.com>
|
2026-06-04 00:14:37 +03:00 |
|
|
Cheng WanandClaude Opus 4.8
|
e485ad6ac1
|
Fix hybrid linear attention dispatch by layer id with draft-worker awareness (#27120)
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
|
2026-06-03 14:11:14 -07:00 |
|
|
Kevin Flansburg
|
7716fa00e0
|
Fix TokenizerManager crash on top_logprobs with tensor values (#26825)
|
2026-06-03 13:55:02 -07:00 |
|
|
|
293816ab14
|
[AMD][MXFP4] Online MXFP4 quantization 1/N - dense and MOE models w. original BF16 weight (#18005)
Co-authored-by: Bowen Bao <bowenbao@amd.com>
Co-authored-by: Colin Zeng <Colin.Zeng@amd.com>
|
2026-06-03 12:55:24 -07:00 |
|
|
Hanming LuandYAMY
|
e0b692600f
|
[Mamba] extra buffer lazy support (#27118)
Co-authored-by: YAMY <74099316+YAMY1234@users.noreply.github.com>
|
2026-06-03 12:42:11 -07:00 |
|
|
Liangsheng YinandAlison Shao
|
ac99794e64
|
Reland spec v2 tree drafting (eagle topk>1) with page_size==1 (#26866) (#26997)
Co-authored-by: Alison Shao <54658187+alisonshao@users.noreply.github.com>
|
2026-06-03 15:40:05 -04:00 |
|
|
Mohammad Miadh Angkad
|
7f706f4cfb
|
[Deps] Bump FI to 0.6.12 and cutedsl to 4.5.2 (#26854)
|
2026-06-03 12:09:18 -07:00 |
|
|
Liangsheng Yin
|
578f232e5e
|
Fix trace_modules gate disabling default trace contexts (#27173)
|
2026-06-03 14:03:42 -04:00 |
|
|
Cheng WanandClaude Opus 4.8
|
45604a0f4a
|
[refactor] Unify CUDA graph runner input buffers behind CudaGraphBufferRegistry (#26742)
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
2026-06-03 10:54:10 -07:00 |
|