Commit Graph

Select branches

Hide Pull Requests

dsv41-pd

dsv41-pd-visioncp

main

Mono Color

• 1d4ee060c2 Support spec v2 tree drafting (eagle topk>1) with page_size==1 (#26866) Liangsheng Yin 2026-06-01 15:37:20 -07:00
• 2fdae94e46 docs: update RTX PRO 6000 deployment snippet (#26968) Baizhou Zhang 2026-06-01 14:34:27 -07:00
• 5700790c05 DeepSeek V4: Support context parallelism with fused MoE (non-DeepEP) (#24947) Yongfei Xu 2026-06-02 05:25:43 +08:00
• 3bce192bd2 [misc] update adaptive spec decoding code owners (#26965) Qiaolin Yu 2026-06-01 14:08:09 -07:00
• 524ba10eda feat: SM120 (Blackwell Desktop) support for DeepSeek-V4 inference (#24692) eeechoandClaude Opus 4.6 2026-06-02 05:05:20 +08:00
• dfa1af99f5 Fix kill_process_tree reap wait crashing on pidfd EINVAL (#26964) Liangsheng Yin 2026-06-01 13:56:59 -07:00
• a0670b5ba3 [SPEC] feat: add adaptive speculative decoding metrics (#25940) 3 people 2026-06-02 04:53:30 +08:00
• 106092123f Update Qwen3-Coder docs_new NVIDIA guidance (#24435) Shu Wangandzijiexia 2026-06-01 15:38:34 -05:00
• da01f2974e [Log] include max_token_num and hidden_dim in FlashInfer workspace init log (#26605) Khoa Pham 2026-06-01 13:26:52 -07:00
• f2beb7bc76 [diffusion] improve: avoid cosmos3 cpu float video postprocess (#26956) Mick 2026-06-02 04:12:01 +08:00
• cb8a103b81 chore: add @pyc96 as codeowner for FrozenKVMTP module (#26953) Khoa PhamandClaude Opus 4.8 2026-06-01 13:06:58 -07:00
• 9a8ab2d22b [diffusion] fix: align cosmos3 text packing with official pipeline (#26950) Mick 2026-06-02 02:07:17 +08:00
• 86afa21ca7 feat: optional caller-supplied mm_hashes on GenerateReqInput (#25300) 3 people 2026-06-01 11:04:37 -07:00
• f6a5a1b59c [RL+VLM] Avoid retokenization drift for pre-tokenized (token-id) VLM requests (#26555) 5 people 2026-06-01 09:58:14 -07:00
• 1988a2c9ea [diffusion] feat: improve cosmos3 serve API support (#26926) Mick 2026-06-02 00:53:39 +08:00
• ed24e3aae8 [diffusion] feat: speed up png image output saving (#26947) Mick 2026-06-02 00:43:02 +08:00
• f59bbef841 Split SWA leaf to one window on insert (#26919) Ke Bao 2026-06-01 23:46:55 +08:00
• d8a5a25c36 Refactor NIXL hicache. Add O_DIRECT support (#25173) Lukas HumbelandClaude Opus 4.7 2026-06-01 17:28:53 +02:00
• 89feb18eb9 [diffusion] feat: allow --dit-cpu-offload with --dit-layerwise-offload (#26925) 3 people 2026-06-01 23:17:53 +08:00
• 693adabff7 Fix Mamba2Metadata dropping has_mamba_track_mask (#26877) Ke Bao 2026-06-01 22:03:59 +08:00
• fd16e05252 [NPU] fix npu profiler (#24835) zhaozx-cn 2026-06-01 21:44:24 +08:00
• 6965fe0eec [sgl] Window-aware LRU refresh for SWA prefix cache in unified cache (#26615) Bi Xue 2026-06-01 04:35:18 -07:00
• 931765e23e Do not cap DeepSeek V4 PD prefill by SWA pool size (#26607) Chi McIsaac 2026-06-01 04:29:45 -07:00
• 1f8d3c7a42 [Speculative] [NPU] Adaptive-SD NPU support (#25644) yiheng 2026-06-01 19:19:58 +08:00
• 1bff7a290f Refactor EAGLE infer tests: shared fixture + kits + overlap matrix (#26871) Liangsheng Yin 2026-06-01 03:55:01 -07:00
• 89410b380b [AMD] Pin compressed-tensors==0.15.0 to fix ROCm nightly build (#26879) Bingxu ChenandClaude Opus 4.8 2026-06-01 17:33:00 +08:00
• 2394dede0e [EPD][Perf] Async image preprocessing and cross-request ViT batching for encode_server (#25669) giang_ng_tr 2026-06-01 15:52:12 +07:00
• bc36231d65 [KDA] Support KDA packed decode (#26586) Yuan Luoandluoyuan.luo 2026-06-01 16:52:01 +08:00
• d078cb72bd [NPU] [DOC] clarify Ascend NPU exclusive supported values for speculative args (#26903) amote-i 2026-06-01 16:40:11 +08:00
• b14fba17d9 [AMD] make bypass-fastfail label also disable within-suite fast-fail (#26909) 4 people 2026-06-01 16:17:59 +08:00
• cdd06011a1 Make unified tree SWA hicache tests faithful to write-through backup (#26870) Ke Bao 2026-06-01 16:08:57 +08:00
• ff642ed936 [MoE] Extend kimi_k2_moe_fused_gate to support 256 experts (MiMo V2 Flash) (#26303) Mengxuan Xiong 2026-06-01 16:03:58 +08:00
• f60710a1d7 [AMD] Fix stage-b-test-large-8-gpu-mi35x-disaggregation-amd : switch CACHE_HOST to a fresh path to fix "No space left on device" (#26905) YC Yen-Ching Tseng 2026-06-01 16:00:58 +08:00
• 1d7e2f6fb8 [NPU] fix normal DeepEP mode num_tokens_per_rdma_rank error caused by none (#22972) ZeyuanChen2000 2026-06-01 15:32:14 +08:00
• 20f47cfe8e fix : add sglang script as entry bin for runtime docker image (#26895) WingEdge777 2026-06-01 15:22:37 +08:00
• afd2d0b2f4 [PP][Bugfix] Handle input_ids assignment in prepare_for_extend (#26883) Shangming Cai 2026-06-01 14:43:50 +08:00
• 53b8378307 Fix weights_checker checksum for 0-dim tensors and multi-GPU (#26863) Lianmin Zheng 2026-05-31 21:27:03 -07:00
• 4b0453f814 [diffusion] CI: infer diffusion test sampling params from task type (#26530) Mick 2026-06-01 11:57:41 +08:00
• a779791b3f Add random-ids dataset, round-robin expert simulation, and kill_process_tree logging (#26862) Lianmin ZhengandJaewon 2026-05-31 20:50:29 -07:00
• 11411aa49d [tokenizer] Surface scheduler load info (num_running_reqs / num_waiting_reqs) in meta_info (#24000) Bruce Changlong Xu 2026-05-31 20:45:35 -07:00
• 3aaf8f115e fix test cases failed in nightly pipeline (#26714) liuxianglong17 2026-06-01 11:33:47 +08:00
• 118465f5b5 [attn backend] Make spec_v2 seq_lens_cpu optional in trtllm_mla backend (#26824) Qiaolin Yu 2026-05-31 20:29:50 -07:00
• 61cc70e8aa Fixed incorrect indexing for slot 0 compatibility (#26481) Chandrakant Khandelwal 2026-06-01 08:16:21 +05:30
• 4d20dc44fc 【NPU】add MiniMax2.5 best practice docs (#26725) shadowxz109 2026-06-01 10:09:12 +08:00
• 1ee189831f [CI] Bump xeon PR test unit tests timeout to 60 minutes (#26682) jundu 2026-06-01 09:13:09 +08:00
• 373cadc92e [bugfix] mooncake store double-tag bug fix (#26569) huangtingwei 2026-06-01 06:15:38 +08:00
• c06220159b [mem_cache][1/N] refactor: split allocator.py into allocator/ subpackage (#26675) shuwenn 2026-05-31 20:31:29 +08:00
• 972fbf7711 Skip flaky mamba extra_buffer disagg test (#26838) Ke Bao 2026-05-31 15:59:03 +08:00
• 585baa97f7 [core] Compute token_type_ids in ForwardBatch.init_new (#26797) Liangsheng Yin 2026-05-31 00:54:49 -07:00
• 1eadb7a173 Fix multi-tokenizer batch request output routing (health stuck at 503) (#26831) ybyang 2026-05-31 15:31:23 +08:00
• 376635c1e3 Fix routed-experts device buffer overflow under DP attention (#26123) Bruce Changlong Xu 2026-05-30 19:11:06 -07:00
• f220c72929 Add periodic KV-canary stats logging and kernel-run-counter health check (#26821) fzyzcjy 2026-05-31 10:00:19 +08:00
• 7dd19ae3d8 Add a sliding-window-attention divergence reporter for the KV-canary (#26820) fzyzcjy 2026-05-31 09:59:28 +08:00
• ae9db7ff4b Add the KV-canary perturb modes and PD-disaggregation e2e tests (#26819) fzyzcjy 2026-05-31 09:59:09 +08:00
• 6be4b32d8d Add token-id verification to the KV-canary (#26818) fzyzcjy 2026-05-31 09:58:51 +08:00
• 0ca610a6df Add real-data KV verification to the KV-canary (#26817) fzyzcjy 2026-05-31 09:58:32 +08:00
• cdee16e144 Add the KV-canary perturb framework for fault-injection self-tests (#26816) fzyzcjy 2026-05-31 09:58:14 +08:00
• 678e73a9ee Add a deterministic token oracle and production write-input assertion (#26815) fzyzcjy 2026-05-31 09:57:43 +08:00
• 268f4c82f1 Add rids/bootstrap-room int-hash plumbing for deterministic per-request identification (#26814) fzyzcjy 2026-05-31 09:57:22 +08:00
• 9c43c3719f Support EAGLE speculative decoding in the KV-canary (#26813) fzyzcjy 2026-05-31 09:57:01 +08:00
• 30a22cc360 Add a periodic full-radix-tree KV-canary sweep (#26812) fzyzcjy 2026-05-31 09:56:42 +08:00
• 27eb139ef7 Add the KV-canary mock-model end-to-end test harness (#26811) fzyzcjy 2026-05-31 09:56:00 +08:00
• e188745ee0 Add KV-canary SWA + DeepSeek-V4 pool support (#26810) fzyzcjy 2026-05-31 09:55:34 +08:00
• 9ecf314970 Add the KV-canary install API and forward-path wiring (#26809) fzyzcjy 2026-05-31 09:55:03 +08:00
• 11391b2a1c Add the KV-canary core: data layer, MHA KV-pool patcher, and per-forward runner (#26808) fzyzcjy 2026-05-31 09:54:24 +08:00
• 736ad1f32a Add the KV-canary plan JIT kernels (#26807) fzyzcjy 2026-05-31 09:53:56 +08:00
• 70e983a2da Add the KV-canary write JIT kernel and reference implementation (#26806) fzyzcjy 2026-05-31 09:53:34 +08:00
• 16950954c6 Add the KV-canary verify JIT kernel and reference implementation (#26805) fzyzcjy 2026-05-31 09:52:52 +08:00
• 8a20c58252 Pull test_utils server-launch boilerplate into reusable helpers (#26804) fzyzcjy 2026-05-31 09:52:15 +08:00
• ce647cbcfd Add a SimplePhaseChecker for execution-phase assertions (#26803) fzyzcjy 2026-05-31 09:51:39 +08:00
• 7f93952f79 Add a debug toggle for selectively reverting PR fixes (#26802) fzyzcjy 2026-05-31 09:50:12 +08:00
• 656e75b798 Add a nullcontext placeholder in the forward path for KV-canary (#26801) fzyzcjy 2026-05-31 09:49:32 +08:00
• bad83ab427 Fix the EAGLE chunked-prefill next-token chain (#26329) (#26800) fzyzcjy 2026-05-31 09:48:30 +08:00
• 45194794d0 Apply gemma's position offset out-of-place instead of in-place (#26799) fzyzcjy 2026-05-31 09:47:50 +08:00
• 13ca55afa3 Make qwen3's set_embed_and_head idempotent (#26798) fzyzcjy 2026-05-31 09:47:07 +08:00
• 9b4be9c574 [core] Compute dimensions/return_pooled_hidden_states in ForwardBatch.init_new (#26779) Liangsheng Yin 2026-05-30 17:38:01 -07:00
• 7ed53d15f3 [config] Recognize custom hybrid SWA models via hf_text_config.is_hybrid_swa (#23988) Bruce Changlong Xu 2026-05-30 08:21:10 -07:00
• e279b0bf72 Optimize large add_constant tensors (#24755) 3 people 2026-05-30 22:25:19 +08:00
• b421e60eed 【NPU】【bugfix】fix server error when mtp unquant (#26389) 3 people 2026-05-30 20:01:19 +08:00
• 282c46133f [Scheduler] Defer prefill input_ids H2D to forward stream, unify resolve via future_map (#25945) Liangsheng Yin 2026-05-30 02:58:13 -07:00
• acd689b407 feat: add SGLANG_RAY_BUNDLE_INDICES for fine-grained Ray bundle index control (#24667) Haichuan Hu 2026-05-30 17:19:50 +08:00
• 90eb894564 [NPU] fix model llava-onevision-qwen2-7b-ov torch compiles error in npu case (#26573) ZeyuanChen2000 2026-05-30 17:10:58 +08:00
• 714bcd84e2 [parallel] Support moe_dense_tp_size == attn_tp_size to share the attention TP group (#23996) Bruce Changlong Xu 2026-05-30 02:10:07 -07:00
• a952e9174f [MLX] Support Qwen3.5 (dense) Model (#25754) 3 people 2026-05-30 17:05:02 +08:00
• 7c5708cba7 [DeepSeek-V4] Add mhc_fused_post_pre kernel (#25976) Qichao LiandQichao Li 2026-05-30 17:04:51 +08:00
• 0d9a2a9de3 [MoE Refactor] Migrate SM90 Cutlass W4A16 to MoeRunner (#26489) Yuan Luoandluoyuan.luo 2026-05-30 17:02:56 +08:00
• c93a559e5a Fix MoE LoRA wrapper exposing moe_runner_config (#26710) huangtingwei 2026-05-30 16:59:15 +08:00
• 3b61a1f935 [Bugfix] Optimize metadata allocation and transfer for mooncake intraNode NVLink (#26707) 3 people 2026-05-30 16:52:01 +08:00
• edfe8d34e8 [CI] ci-coverage-overview: schedule + manual only, include XPU/MUSA/multimodal_gen (#26619) Alison Shao 2026-05-30 01:39:18 -07:00
• b4bf489dea ci: allow /rerun-test to dispatch nightly/weekly CUDA tests (#26624) Alison Shao 2026-05-30 01:35:01 -07:00
• 716e670d3d [bugfix]: size CuteDSL MoE allgather buffers for the worst-case forward (#26696) Jimmy Shong 2026-05-30 00:27:20 -07:00
• 7662210406 [UnifiedTree]: Support eviction priority (#26549) Zhangheng 2026-05-30 15:19:43 +08:00
• c048ebd10d [Hicache]: skip flaky test (#26764) Zhangheng 2026-05-30 15:16:12 +08:00
• 02aeed5387 [NPU] DFlash Speculative Decoding Support NPU (#23122) gjsheu 2026-05-30 15:13:59 +08:00
• fe4b29d391 [Bugfix] Fix Ascend NPU CP attention for batch size > 1 (#26705) AndyLi429 2026-05-30 15:07:39 +08:00
• 23a825c694 [DOC] [NPU] add qwen3.5-397b best practice to doc_new (#26709) silencejade 2026-05-30 14:20:24 +08:00
• 1c79015434 Drop dead ScheduleBatch return_routed_experts/return_indexer_topk fields (#26760) Liangsheng Yin 2026-05-29 23:16:09 -07:00
• 6f1c9fc77b [RL] Fix crash when the reqs in a batch have a mix of return_routed_experts = True and False. (#26423) 4 people 2026-05-29 20:46:10 -07:00
• 804f01a823 Allow PR test and lint workflows to trigger on non-main bases (#26721) fzyzcjy 2026-05-30 10:22:59 +08:00
• 1f850e67f2 [core] Fix crashes on the gpu_only spec_v2 path (#26738) Liangsheng Yin 2026-05-29 19:12:03 -07:00