Commit Graph

6985 Commits

Author SHA1 Message Date
Baizhou Zhang
39008955ff Revert "[AMD][MORI] Fix MTP crash with FP4/FP8 dispatch and add NEXTN dispatch env vars." (#20602) 2026-03-14 12:12:42 -07:00
Xiaoyu Zhang
5ab2cfe9a8 [Diffusion] Clean upstream fa3 in hopper (#20576) 2026-03-14 23:41:23 +08:00
Yuan Luo
22e67876d6 [Omni] Optimize AudioEncoder for Qwen3_Omni_Thinker (#18185)
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
2026-03-14 23:00:17 +08:00
Ratish P
574aa2d723 [diffusion]: remove stale offload-manager cleanup in denoising stage (#20587) 2026-03-14 22:56:57 +08:00
Xiaoyu Zhang
25e38216b6 [kernel slimming] Clean many useless sgl-kernel deprecated kernels (#20277) 2026-03-14 16:45:54 +08:00
Mohammad Miadh Angkad
75a7879fd4 [Model] Support Nemotron 3 Super NVFP4 (#20407) 2026-03-14 00:56:26 -07:00
SoluMilken
c95dc88f86 [CI] migrate ascend-gptq from test/srt to test/registered (#19628) 2026-03-14 00:28:57 -07:00
Xiaoyu Zhang
f9e4221b71 [Diffusion] add mova and hunyuanvideo to perf skills (#20563) 2026-03-14 13:49:50 +08:00
Shangming Cai
99a3b25c9b [PP] Fix recv tensor dict potential race condition (#20341)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-03-14 13:35:01 +08:00
Xinyuan Tong
c330b687a1 [Bugfix] Fix GLM-4.6V vision regression in glm4v_moe and glm_ocr (#20463) 2026-03-13 21:48:28 -07:00
ziruiliu
dfd0a77a9a [bugfix] Add prev_prefix_len parameter to HiMambaRadixCache's _insert_helper() (#20539) 2026-03-14 09:54:14 +08:00
Duyi-Wang
0eea80bc00 [AMD][MORI] Fix MTP crash with FP4/FP8 dispatch and add NEXTN dispatch env vars. (#20453) 2026-03-13 14:03:17 -07:00
YC Tseng
c37ef7f18b [AMD] diffusion refactor: move ROCM VAE optimization to Platform abstraction (#20496) 2026-03-13 13:10:05 -07:00
Simo Lin
654fc02cf1 [gRPC] Extract gRPC servicer into standalone package (#20478)
Signed-off-by: Simo Lin <linsimo.mark@gmail.com>
2026-03-13 09:13:29 -07:00
Xiaoyu Zhang
be7a0311a0 [Diffusion] Fix and validate diffusion skills benchmarking/profiling workflow (#20528) 2026-03-13 21:11:37 +08:00
Leon Gao
b1246c50f8 Fix chunked prefill and KV cache leaks for streaming sessions (#20476)
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: hnyls2002 <lsyincs@gmail.com>
2026-03-13 02:36:55 -07:00
Ke Bao
287dc12b05 Fix hicache log metrics (#20504) 2026-03-13 16:29:58 +08:00
Baizhou Zhang
f8668d9e78 [Fix] Add fallback for flashinfer allreduce fusion (#20384) 2026-03-13 01:24:55 -07:00
Mick
b638b25b22 [diffusion] UX: suppress excessive logging from httpx and httpcore (#20452) 2026-03-13 14:43:09 +08:00
seungrokj
9c8777c80f [AMD][Qwen3.5] aiter a8w8 gemm configuration (#19826)
Signed-off-by: seungrokj <seungrok.jung@amd.com>
Co-authored-by: HaiShaw <hixiao@gmail.com>
2026-03-12 23:23:58 -07:00
Antonin Vidon
63ecdcbb18 Expose async LoRA interface to Offline Engine (#18636) 2026-03-12 23:09:47 -07:00
StonyPort
d4e68ead1d [quant] Ignore FP8 quantization layers (#20340)
Co-authored-by: qiuxuan.lzw <qiuxuan.lzw@alibaba-inc.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
2026-03-13 13:59:39 +08:00
Xiaoyu Zhang
e00328d1e5 [Diffusion] Opt qwen-image-edit with fuse_residual_layernorm_scale_shift_gate_select01_kernel (#20395)
Co-authored-by: Yihan Chen <yingluosanqian@gmail.com>
2026-03-13 13:15:22 +08:00
hzh0425
197f807134 [RadixTree][7/N Refactor]: Refactor mamba radix tree, release dup kvcache in insert func (#19429) 2026-03-13 12:28:32 +08:00
Liangsheng Yin
f605612b87 [HTTP] Fix /GET HTTP route when ollama endpoint is not set. (#20494) 2026-03-12 20:54:32 -07:00
Xiaoyu Zhang
7ecf07b8f4 [jit_kernel] Temporarily Skip Flaky JIT Kernel GDN Test and Add PR Label (#20436) 2026-03-13 09:34:22 +08:00
Pai Liu
65dd08153d Fix Test* mixin classes being collected as standalone pytest tests (#20417) 2026-03-12 18:18:45 -07:00
LinyuanLi
9865f11421 [bugfix] fix bug when enable prefill delay and DP (#20134) 2026-03-13 08:51:09 +08:00
zzhpro
c21ddbc785 [Minor] fix type annotations and invalid method calls in constrained … (#20132) 2026-03-12 16:42:46 -07:00
YC Tseng
78a467c74a [AMD] [diffusion] feat: enable AITer GroupNorm for VAE decode on ROCm (#20170)
Co-authored-by: HaiShaw <hixiao@gmail.com>
2026-03-12 16:38:19 -07:00
Kangyan-Zhou
f5a4a5429f Revert early HTTP port reservation (#17754, #19805) (#20468)
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-12 16:17:33 -07:00
Ethan (Yusheng) Su
af2807e146 [LoRA][I] Add MOE LoRA JIT alignment kernel and tests (#19710)
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Jonah Bernard <96398205+Jonahcb@users.noreply.github.com>
2026-03-12 12:23:46 -07:00
Yuhao Yang
a57a44739f [diffusion] deps: upgrade diffusers from 0.36.0 to 0.37.0 (#20318) 2026-03-12 19:17:28 +08:00
kk
318a40fdfb [Bug-fix] Fix gpu fault when run the test with dp-attention-enabled and max-concurrency is over 256 (#20399)
Co-authored-by: wunhuang <wunhuang@amd.com>
2026-03-12 02:32:03 -07:00
Ratish P
4e5ca92249 [diffusion]: clear file-path-only outputs on all ranks to prevent TP GPU memory skew (#20353) 2026-03-12 17:29:09 +08:00
jacky.cheng
1e2983c98e [AMD] Fix FP8 assertion failure in aiter MLA decode by falling back to self.k_scale (#19935) 2026-03-12 01:48:51 -07:00
roikoren755
067353f67b [Test] Refactor KL divergence and prefix cache branching to kits (#19715) 2026-03-12 16:11:59 +08:00
0xNullPath
46b558445d Fix default_max_tokens compute error in responses api when mtp is opened (#18932) 2026-03-12 16:00:48 +08:00
Hexq0210
dd82678b2d [NPU] Support mamba cache transfer for NPU (#20364) 2026-03-12 12:49:21 +08:00
Mook
abc672e717 [Benchmark] use flashinfer bench_gpu_time instead of triton do_bench (#20305) 2026-03-12 04:04:30 +00:00
Ke Bao
ae7c2397b9 Fix FA3 swa spec pg_size > 1 (#20369) 2026-03-12 11:42:01 +08:00
Yuan Luo
649d6f2bc8 [GDN] Change Attention State Layout from [N, HV, K, V] to [N, HV, V, K] (#20283) 2026-03-12 10:53:12 +08:00
huangtingwei
8787cf4566 Fix the scope of io_backend in NSATokenToKVPoolHost (#20327) 2026-03-12 10:33:11 +08:00
Vedant V Jhaveri
9b55a98a67 perf(qwen3_5): replace einops rearrange with torch.flatten in GatedDe… (#20386) 2026-03-12 09:51:27 +08:00
Vedant V Jhaveri
25bd83033d Enable Piecewise CUDA Graph for NemotronH Hybrid (Mamba+Attention) Models (#19903) 2026-03-12 09:16:38 +08:00
fy
677e446e51 [NPU] Convert cu_window_seqlens to CPU for npu_flush_attention_unpad operator (#20328) 2026-03-12 09:08:43 +08:00
Hubert Lu
67f02681c9 [AMD] Support speculative decoding v2 for aiter backend on ROCm/HIP (#17450)
Co-authored-by: kkHuang-amd <wunhuang@amd.com>
Co-authored-by: HaiShaw <hixiao@gmail.com>
2026-03-11 17:01:01 -07:00
shuwenn
acab24a76a fix: gracefully abort last request in retract_decode on OOM (#19881) 2026-03-11 15:13:03 -07:00
doujiang24
88d2fc19b1 feature: support X-Data-Parallel-Rank header to specific dp-rank. (#19832)
Signed-off-by: doujiang24 <doujiang24@gmail.com>
2026-03-11 14:53:33 -07:00
Shangming Cai
af4c28904d [PD] Fix the infinite loop in deocde resolve_pending_reqs (#20371)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-03-11 14:11:19 -07:00