Commit Graph

Select branches

Hide Pull Requests

bot/update-est-time-20260928

dsv41-pd

dsv41-pd-visioncp

main

Mono Color

• 4ac6fa0d87 [diffusion] fix: fix loading multiple ckpts with different precision for a same module (#22360) Mick 2026-04-09 02:44:19 +08:00
• a5ed507a16 [refactor] [asr] add transcription adapter for extensible ASR models support (#22181) Yihao Wang 2026-04-08 10:19:37 -07:00
• ae8da14ea3 [fix] [whisper] ensure inputs are moved to the correct device before processing. (#22293) Yihao Wang 2026-04-08 08:45:42 -07:00
• b5b2dbe05f [Diffusion] Add diffusion NVFP4 scaled-mm correctness test (#22127) Xiaoyu ZhangandMick 2026-04-08 22:07:24 +08:00
• ea119adc90 Refactor auto benchmark unit tests and fix CI bug (#22270) Xiaoyu Zhang 2026-04-08 21:54:41 +08:00
• 33c9cc8994 [NPU] fix qwen3.5 video processor (#22266) zhaozx-cn 2026-04-08 21:13:29 +08:00
• 931dbceadc [CI] Set RUNAI_STREAMER_MEMORY_LIMIT=0 for stage-b-test-1-gpu-small (#22346) Alex NailsandClaude Opus 4.6 2026-04-08 02:23:35 -07:00
• 2ad5e6df12 [CI] Relax gpt-oss 4GPU accuracy threshold from 0.60 to 0.58 (#22237) Alison ShaoandAlison Shao 2026-04-08 02:20:23 -07:00
• 413913763f fix: wrap _import_static_state in inference_mode to fix resume on Blackwell (#21035) Fergus 2026-04-08 10:03:39 +01:00
• 79c82c5c42 [HiCache] Fix write_backup return type when parent not backed up (#22185) Vladislav Nosivskoyandhzh0425 2026-04-08 11:42:57 +03:00
• 712c8c5051 [Score API] Add SequenceClassification Model support (#22118) Sundara Raman Ramachandran 2026-04-08 01:30:58 -07:00
• 213af1d4f7 Add CI tests for GLM-5 (#22285) Baizhou Zhang 2026-04-08 01:05:36 -07:00
• c3c13dd5e3 [diffusion] fix: make warmup image initialization rank-safe (#21817) HuangJi 2026-04-08 15:51:09 +08:00
• de0cfed159 [AMD] Fix DLPack Error in Aiter flydsl GEMM by Detaching MoE Gate Weight (#22262) Bingxu Chenandbingxche 2026-04-08 14:42:10 +08:00
• cd373667cd [Bugfix] [NPU] Qwen3.5 with quantization fix (#21692) Артем Савкин 2026-04-08 09:15:48 +03:00
• db60a620db [AMD] Add GLM-5-FP8 nightly performance benchmarks for MI30x and MI35x (#21710) Michael 2026-04-07 22:43:14 -07:00
• 729b74d8dd [AMD] Fix GLM-5 fp8 KV quant path dispatch on MI300 (#22314) Thomas Wang 2026-04-08 12:16:02 +08:00
• 36f05810c9 [CI] Move manual-only nightly tests out of test/registered/ (#22298) Alison ShaoandAlison Shao 2026-04-07 21:03:52 -07:00
• 4e4b4ac153 [NPU] enable index Cache for npu (#21502) yuefeng Wu 2026-04-08 11:45:17 +08:00
• 493ec91cbe [CI] Fix stage-b-test-1-gpu-large (0) timeout by reordering LoRA tests and using tokenizer from cache (#22292) Alex NailsandClaude Opus 4.6 2026-04-07 20:00:44 -07:00
• 86e4542f35 Use dedicated runner label for deepep 8-GPU tests (#22309) Alison ShaoandAlison Shao 2026-04-07 19:58:54 -07:00
• 1c5c6dad5e [tiny] Fix TOCTOU race in pause-aware weight update locking (#22304) 3 people 2026-04-07 18:54:28 -07:00
• eca62ab8f4 UX: clean loggings (#22174) Mick 2026-04-08 09:46:38 +08:00
• 117508dcd7 Switch eagle_infer_beta to EAGLE3 (#22303) Qiaolin YuandLiangsheng Yin 2026-04-07 18:43:48 -07:00
• 6c2a759a04 [fix] Fix writer lock deadlock in update_weights_from_ipc during pause_generation (#22290) maocheng23andClaude Opus 4.6 2026-04-07 18:32:56 -07:00
• 8c3d80eabe Only upload CUDA coredumps on test failure (#22301) Liangsheng Yin 2026-04-07 18:07:28 -07:00
• dd73e9a62e Revert "[CI] Update nightly test models for H200/B200 (#22288)" (#22297) Kangyan-Zhou 2026-04-07 17:04:06 -07:00
• f6fc39569a [CI] Migrate mgsm_en eval to gsm8k to remove openaipublic dependency (#21931) 3 people 2026-04-07 16:29:20 -07:00
• 7546d04c81 [NVIDIA] Enable FP4 flashinfer trtllm routed moe (#21240) Trevor Morris 2026-04-07 16:16:29 -07:00
• 0e2a0260a1 Add fast-fail to multimodal-gen CI (#22284) Liangsheng Yin 2026-04-07 15:56:12 -07:00
• e6652309c4 [CI] Update nightly test models for H200/B200 (#22288) Kangyan-ZhouandClaude Opus 4.6 2026-04-07 15:44:52 -07:00
• 671fe73961 Reduce unnecessary kernels and copies in the NSA indexer (#22232) Thomas Wang 2026-04-08 06:37:08 +08:00
• f08726fd56 [Feature] Add DFLASH speculative decoding support (#22077) 6 people 2026-04-07 17:48:51 -04:00
• e14876742a [AMD] Fix test_kimi_k25_mxfp4.py : stage-c-test-large-8-gpu-amd-mi35x (linux-mi35x-gpu-8, 1) (#22188) YC Yen-Ching Tseng 2026-04-08 04:48:37 +08:00
• cc35714b03 [tiny] migrate /get_server_info; print accept length in accuracy tests (#22282) Liangsheng Yin 2026-04-07 13:08:35 -07:00
• 1a8eb890f6 Kernels community fa3 (#20796) Rain Jiang 2026-04-07 12:48:44 -07:00
• 0c204fbd57 [HiSparse] Optimize the scheduling of decode backup. (#21932) 3 people 2026-04-08 01:34:58 +08:00
• 6131fb5882 [NPU] enable mla prepare fused kernel only when being mla attn (#22024) khalilzhk 2026-04-08 00:49:16 +08:00
• be42fbbbd7 Support HTTP2 server (#21700) Ke Bao 2026-04-08 00:42:52 +08:00
• ec5742f4ab fix: Auto-correct page_size for Mamba no_buffer radix cache mode (#20538) shuwenn 2026-04-08 00:19:31 +08:00
• 727a182067 [Mamba] eliminate D2H if tracking mamba states (#20522) Henson-Zh-Aliandhzh0425 2026-04-08 00:17:26 +08:00
• 5ae00ecd48 [Disagg][NIXL] Support Mamba state slice transfer for heterogeneous TP (Step 2/2 for Qwen3.5) (#22240) YAMY 2026-04-07 08:47:31 -07:00
• e7bc23cdab [diffusion] CI: fix consistency check (#22251) Mick 2026-04-07 23:43:18 +08:00
• fae90abf6e Move ring test to nightly (#22267) Ke Bao 2026-04-07 21:56:39 +08:00
• 233f3e31bf fix(pcg,mm): fix zeroing of input_embeds when replay PCG (#22229) Yujun Dong 2026-04-07 20:33:17 +08:00
• 98f38b14df Add registration API for external linear attention backend (#21983) Xingyu Liu 2026-04-07 02:47:40 -07:00
• 490fa9fa44 [Perf] Restore torch.compile fusion for topk postprocessing (#21771) Nicolas Castet 2026-04-07 03:38:38 -05:00
• 3d3a32c0b9 [HiSparse]: Add readme docs for HiSparse Feature (#22238) Zhangheng 2026-04-07 15:39:24 +08:00
• 3148742ddb [Disagg][NIXL] Fix heterogeneous TP KV transfer for non-MLA models (same logic with mooncake, Step 1/2 for Qwen3.5 support) (#22145) YAMY 2026-04-06 23:52:02 -07:00
• ba78f6e0ef [AMD] Add Qwen3.5-397B FP8 nightly perf benchmarks for MI30x and MI35x (#21669) Michael 2026-04-06 23:46:00 -07:00
• 3f7dfba419 fix qwen2_5_math_rm_72b (#21295) amote-i 2026-04-07 14:36:57 +08:00
• f6e85676b5 model: support qwen3-asr (#22073) Aditya SharmaandXinyuan Tong 2026-04-07 10:57:05 +05:30
• a757c1e3fb [Apple Silicon] [MLX] Add mlx and mlx-lm dependencies (#22162) Chang Min BarkandR0CKSTAR 2026-04-06 23:36:43 -04:00
• 2813cb6d9a [New Model] Gemma 4 (#21952) 7 people 2026-04-07 04:24:44 +01:00
• be0277f9a0 [Spec][Ngram] Add output-as-corpus accept length benchmark for external SAM (#22199) Liangsheng Yin 2026-04-06 19:09:52 -07:00
• 73fc87a74f fix(grok): adapt huihui-ai/grok-2 (#21522) jianzhao-xuandJianzhao Xu 2026-04-07 10:04:41 +08:00
• 494bb86169 Cache sub-objects in __getitem__ to ensure identity stability (#22184) Lianmin ZhengandClaude Opus 4.6 2026-04-06 18:53:38 -07:00
• ef2d4013d7 [diffusion] CI: add consistency test (#15236) Prozac614anddaiweitao 2026-04-07 09:50:23 +08:00
• e4b1366a46 [Spec][Ngram] Support multiple SAMs with dynamic HTTP API (#22203) Liangsheng Yin 2026-04-06 18:49:22 -07:00
• 49cb7d546e Move hash utils out of hicache_storage to break CUDA import chain (#22214) Liangsheng Yin 2026-04-06 18:16:40 -07:00
• 5e2b0f860c [diffusion] perf: replace Conv3d with reshape + F.linear in PatchEmbed (#21014) AichenF 2026-04-07 09:12:59 +08:00
• ae38b24cc3 [NPU] Support dp-attention for MiniMax2.5 (#20919) shadowxz109 2026-04-07 08:55:37 +08:00
• 5cc246e095 Fix extra calls to get_numa_node_if_available to clean up logs (#21781) Trevor Morris 2026-04-06 16:18:40 -07:00
• 56266de624 [CI] Add basic unit test for Minimax-M2.5 (#21792) Trevor Morris 2026-04-06 15:48:33 -07:00
• 6f1412f4f5 [CI] Relax transformers MMLU threshold from 0.65 to 0.64 (#22210) Alison ShaoandAlison Shao 2026-04-06 15:32:09 -07:00
• a80961333b Clean up req_time_stats: reduce overhead and simplify (#22186) Lianmin Zheng 2026-04-06 14:20:51 -07:00
• 0bc4e0ea75 [misc] update CI_PERMISSIONS.json (#22207) Qiaolin Yu 2026-04-06 14:11:49 -07:00
• 93f38fe410 tiny fix chain-style multi layer eagle comments (#22206) Qiaolin Yu 2026-04-06 13:49:03 -07:00
• 8f337682bd [sgl] potential chained spec v2 fixes (#22041) 3 people 2026-04-07 04:38:04 +08:00
• 7f2fcc0b08 [VLM]: allow Qwen3.5 models for encoder disaggregation (#21849) Ratish P 2026-04-06 23:37:24 +05:30
• d72f58d1c1 [Qwen3-Specv2]: Fix flaky ci (#22194) Zhangheng 2026-04-07 00:40:44 +08:00
• b95324d19c Update test coverage report (#22190) Ke Bao 2026-04-06 20:59:38 +08:00
• 9ca2ae1c6c Update test skills and guide (#22189) Ke Bao 2026-04-06 20:30:25 +08:00
• 3178f3959f Align incremental streaming logprobs with streamed output tokens (#21583) Aurick Qiao 2026-04-06 00:30:02 -07:00
• 12272b6791 [Spec][Ngram] 6/N: Load an external corpus and construct a Suffix Automaton (#21425) Khoa Pham 2026-04-06 00:11:14 -07:00
• b311db2e49 [Doc] Fix and improve DeepSeek V3.2/GLM-5 documentation (#22179) Mohammad Miadh Angkad 2026-04-06 14:26:42 +08:00
• 6de2ff2a80 [Spec][Ngram] Followup fixes for MatchState incremental advance (#22180) Liangsheng Yin 2026-04-05 23:04:28 -07:00
• dc125afffb Add staging buffer CI test and documentation for heterogeneous TP (#21921) YAMYandShangming Cai 2026-04-05 23:00:20 -07:00
• b2008bf9e0 [Spec][Ngram] 5/N: Store and advance anchor match state across decode steps (#21243) Khoa Pham 2026-04-05 22:21:05 -07:00
• 82c41a2d9e [diffusion] model: support LTX2.3 (#22111) Mick 2026-04-06 12:26:30 +08:00
• f407461ec8 Tiny fix trtllm_fp8_per_tensor_scale_moe_wrapper router_logits dtype (#22006) Qiaolin Yu 2026-04-05 21:11:45 -07:00
• c4240218cb Fix ut module importing (#22176) Ke Bao 2026-04-06 11:53:58 +08:00
• e835601fb7 Cache gfx95 quant format detection in DeepseekV2DecoderLayer (#22143) Lianmin Zheng 2026-04-05 20:20:54 -07:00
• 52801ff20c [sgl] two potential spec_v2 bug fixes (#21589) Bi Xueandyilian49 2026-04-05 19:41:43 -07:00
• 682ddf6416 [CI] Add unit tests for srt/utils/auth.py (#21400) Lidang Jiang 2026-04-06 10:30:46 +08:00
• 30f5b87608 [CI] Add unit tests for function_call detectors (hermes, llama32, mistral) (#21399) Lidang Jiang 2026-04-06 10:29:55 +08:00
• 8d8aca8f26 [Test] Add unit tests for srt/tokenizer/tiktoken_tokenizer (#21107) ROSINE HE 2026-04-05 22:23:19 -04:00
• 2f00e42555 [diffusion] CI: apply diffusers backend in lora case (#22157) Prozac614anddaiweitao 2026-04-06 10:14:35 +08:00
• 41c7c97ff3 fix hisparse LRU policy (#22170) 3 people 2026-04-05 18:47:58 -07:00
• 93109cc89b [Fix] Fix setuptools-scm version resolution for rc tags (#22165) 3 people 2026-04-05 16:55:32 -07:00
• 30ba1f78b0 Hisparse Minor Fix (#22131) 3 people 2026-04-05 16:15:47 -07:00
• 20ee59bcfc [Misc] Remove unused cu13 docker release workflow (#22167) Baizhou Zhang 2026-04-05 16:01:02 -07:00
• 596c34ee04 Update ci_auto_bisect.py to have streak 1 so that all failures will b… (#22161) Kangyan-ZhouandClaude Opus 4.6 2026-04-05 10:39:19 -07:00
• 75ab75d027 Fix create_grammar_backend test calls with think_end_id (#22158) Shangming Cai 2026-04-06 00:54:35 +08:00
• 5dd2c243eb fix: TRT-LLM MHA CUDA illegal address with EAGLE v2 + DP attention (#21649) 3 people 2026-04-05 09:41:14 -07:00
• c5fa364b80 [Hotfix] Fix router gemm on sm103 (#22134) Baizhou Zhang 2026-04-05 09:33:14 -07:00
• f6c9072e42 [SpecV2]: Reopen kl accuracy test for qwen3 + SpecV2 (#22104) Zhangheng 2026-04-05 23:26:40 +08:00
• 51b276de74 [BugFix][RadixTree]: Fix backup invariant violation in Hi-MambaRadixTree (#22062) 3 people 2026-04-05 23:19:50 +08:00
• dccb11881f [PD] Fix staging warmup for GQA prefill decode different tp (#22153) Shangming Cai 2026-04-05 23:13:06 +08:00
• 3a4f4cbc52 DEBUG: reproduce flaky test_load_weights_from_remote_instance (#22150) Liangsheng YinandShangming Cai 2026-04-05 07:46:36 -07:00