Commit Graph

5869 Commits

Author SHA1 Message Date
Yingchun Lai
a5bbcda968 fix: prefer to use max_completion_tokens rather than max_tokens (#17516) 2026-01-21 20:22:20 -08:00
Baizhou Zhang
3373545b9f [HotFix]Fix dtype mismatch in nsa indexer on AMD device (#17518) 2026-01-22 12:17:02 +08:00
Piotr Mazurek
d6e2b88288 Add Liquid Foundation Model (LFM2) (#16890) 2026-01-22 11:11:20 +08:00
cen121212
0c2993eed0 Optimize Qwen3-VL video memory usage (#16366) 2026-01-22 09:10:08 +08:00
Xiaoyu Zhang
590969ee9c [Diffusion] Support select fa2 backend in hopper (#17514) 2026-01-22 08:23:53 +08:00
Lianmin Zheng
b74a57a8d9 [Auto Sync] Update detokenizer_manager.py, io_struct.py, mu... (20260120) (#17442)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Wangfan Fu <wangfan@x.ai>
2026-01-21 14:48:32 -08:00
DarkSharpness
95f59c13fd [Chore] include all jit files in building packages (#17493) 2026-01-21 14:48:02 -08:00
Lingjun Wen
cf89351691 [new-model] Add support for Cohere2ForCausalLM behind Command-A and Command-R Models (#16927) 2026-01-21 12:28:33 -08:00
Lianmin Zheng
1fdf5cac39 [Auto Sync] Update environ.py, fp8.py (20260121) (#17486)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Binyao Jiang <byjiang1996@gmail.com>
2026-01-21 12:04:09 -08:00
Jacob Gordon
cda43ffa4d ci: avoids duplication of codespell config (#17519) 2026-01-21 12:02:37 -08:00
Yunmeng
390898545e [Misc] Fix argument help string formatting (#17416) 2026-01-21 09:25:32 -08:00
Ke Bao
d725487dc8 Disable swa memory for gpt-oss with spec (#17517) 2026-01-22 01:04:19 +08:00
JiaruiChang5268
a95c9f5b81 [NPU] Remove paged attention & Change fia to default attention (#17394)
Co-authored-by: Liwansi <62291011+Liwansi@users.noreply.github.com>
Co-authored-by: chenxu214 <justin_cc2025@163.com>
Co-authored-by: chenyang08056032 <chenyang08056032@163.com>
2026-01-21 23:58:25 +08:00
Xiaoyu Zhang
19089aa431 [Diffusion] Refactor diffusion is_cuda check (#17498) 2026-01-21 23:02:24 +08:00
Qiaolin Yu
4f6f5d25c8 Support fa4 decoding (#16034) 2026-01-21 22:54:02 +08:00
b8zhong
2ff0880a0e [Fix] GLM 4.7 + NVFP4 + MTP (#17166) 2026-01-21 21:34:18 +08:00
Zhu Yuhua
2c1b164a92 [diffusion] improve: skip negative prompt encoding when guidance_scale <= 1.0 or negative_prompt is None (#16919)
Signed-off-by: zhuyuhua-v <yuhzhu@amd.com>
2026-01-21 20:01:55 +08:00
strgrb
bcc6d84f93 Use fused_sigmoid_gating_delta_rule_update_kernel for KDA (#17108) 2026-01-21 19:24:29 +08:00
Ke Bao
a618202fc7 Tiny refine swa kv cache free (#17417) 2026-01-21 19:20:05 +08:00
siyu
7520b92927 Support EPD error handling (#16670)
Co-authored-by: ZhengWG <zwg0606@gmail.com>
2026-01-21 18:47:00 +08:00
Fan Lin
e7224e9681 [diffusion] fix: fix the LoRA weights mismatch caused by weights packing (#17355) 2026-01-21 18:17:16 +08:00
HuangJi
e776239afd [diffusion] feat: support SageSparseLinearAttention attention backend (#17399) 2026-01-21 18:13:51 +08:00
Yi Zhang
1b97fa769b [BUGFIX] fix value oom in radix tree (#17400) 2026-01-21 17:12:57 +08:00
Yi Zhang
236772c0e1 [RadixTree][2/N Refactor]: swa cache init tiny refactor (#17397) 2026-01-21 15:48:30 +08:00
Sam Shleifer
0d49b13fdd Fix circular import in quantization modules (#17372) 2026-01-21 15:47:09 +08:00
blahblah
0a7a2017a0 [diffusion] refactor: refactor and simplify teacache for cachabledit and wanvideo (#16396)
Co-authored-by: Brain97 <Brain97@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Mick <mickjagger19@icloud.com>
Co-authored-by: blahblah <blahblah>
2026-01-21 15:42:45 +08:00
Baizhou Zhang
8251a74d5f [Tiny] Backward compatibility for fp4 gemm flags (#17466) 2026-01-21 14:34:40 +08:00
Baizhou Zhang
a54d75bf2e [Fix] Set fa3 as default MHA backend on Hopper (#17425) 2026-01-21 13:54:09 +08:00
Fan Lin
54a821794e [diffusion] fix: fix the bug of output_path not taking effect when generate (#17293) 2026-01-21 11:39:51 +08:00
Yinghai Lu
aea57b33c6 [scheduler] Clear MM data of finished batch (#17251)
Co-authored-by: Kangyan-Zhou <zky314343421@gmail.com>
2026-01-20 17:45:38 -08:00
Binyao Jiang
6092721594 [Piecewise] Fix PCG issue for multimodal and embedding model that wraps language_model (#17290) 2026-01-20 14:06:06 -08:00
Binyao Jiang
38c233fd04 [Piecewise] Support PCG weak_ref_tensor cuda kernel on AMD (#17291) 2026-01-20 14:05:32 -08:00
Lianmin Zheng
20ed3822bb [Auto Sync] Update piecewise_cuda_graph_runner.py (20260119) (#17313)
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Binyao Jiang <byjiang1996@gmail.com>
2026-01-20 14:05:05 -08:00
Baizhou Zhang
6ea491e439 Overlap shared experts with deepep dispatch for single batch overlap on Blackwell (#17289) 2026-01-21 02:56:55 +08:00
GuoweiWangU
16802fb6b2 [FIX] fix mambaish model pp kv cache compute (#17334) 2026-01-20 23:59:05 +08:00
Ke Bao
d97066d209 Use attn_tp_group for all reduce in token embedding (#17403) 2026-01-20 23:56:21 +08:00
Ke Bao
ce2d686e94 Disable swa memory for trtllm-mha backend (#17429) 2026-01-20 23:55:00 +08:00
Qiaolin Yu
76b06bee03 [New Model] GLM4.7-Flash (#17247)
Co-authored-by: zRzRzRzRzRzRzR <2448370773@qq.com>
Co-authored-by: JustinTong0323 <justinning0323@gmail.com>
Co-authored-by: Xinyuan Tong <xinyuantong.cs@gmail.com>
Co-authored-by: Baizhou Zhang <sobereddiezhang@gmail.com>
2026-01-20 23:44:16 +08:00
BingjiaWang
612026ad2c [OPT] DeepSeekV3.2: optimize indexer weight_proj-mma performance (#17205) 2026-01-20 23:09:44 +08:00
Baizhou Zhang
55c616427d Add flag that enables NCCL mlp sync batch for overlap scheduler (#17288) 2026-01-20 23:06:55 +08:00
Ke Bao
91a4cd8648 Update default attn backend to fa3 for mimo model (#17419) 2026-01-20 21:22:27 +08:00
ympcMark
f7a5e425c3 [3/N] Achieve fault tolerance at the DP level (#11657)
Co-authored-by: UNIDY <unidy2002@outlook.com>
Co-authored-by: Hank Han <hanhan7630@outlook.com>
2026-01-20 18:47:08 +08:00
Yuan Luo
e6b7c04947 [Kimi-Linear] Refactor kimi-linear gate calculation to avoid duplicated code (#17160)
Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>
2026-01-20 14:29:24 +08:00
Thomas Wang
6988a0f570 Disable mla persistent kernel when not using fp8 kv_cache (#17327) 2026-01-19 20:29:54 -08:00
Ratish P
c560e14216 [diffusion] fix: enforce 16-pixel alignment for flux2 to prevent shape mismatch crash (#17302) 2026-01-20 12:03:33 +08:00
Shangming Cai
71cb9d0302 [PD] Fix PP dynamic chunking with DP attention (#17339)
Signed-off-by: Shangming Cai <csmthu@gmail.com>
2026-01-20 11:27:13 +08:00
shuwenn
8fb45523f3 feat: support bitsandbytes quantization algorithm (#15325) 2026-01-19 18:36:56 -08:00
Raghav Ravishankar
84aef37808 fix function calling for Trinity (#17364) 2026-01-19 18:33:17 -08:00
Aurick Qiao
09a9d214f7 Pipe customized_info through CudaGraphRunner output (#17088) 2026-01-19 15:48:49 -08:00
Hudson Xing
c1282da236 fix(ci): apply MMMU retry logic to all affected test files (#17329) 2026-01-19 10:00:29 -08:00