Commit Graph

Select branches

Hide Pull Requests

cch/megamoe-rc-v0.1

Mono Color

• 4d64f15086 [mode;-gateway] add back fixes of incorrect metrics after worker removal (#15624) fzyzcjy 2025-12-24 10:50:03 +08:00
• aef7ca7cf2 Raise the accept length bar in dpsk-r1-fp4 spec decoding tests (#15705) Qiaolin Yu 2025-12-23 18:38:02 -08:00
• fe712aa3df [bug fix] fix hicache jit kernel (#15177) Xuchun Shang 2025-12-24 10:31:21 +08:00
• 846953d9f1 [model-gateway] Add tokenize/detokenize HTTP endpoints and tokenizer management (#15702) Simo Lin 2025-12-23 17:32:07 -08:00
• 5c64a20da7 Update MiniMax-M2 ToolCall and add MiniMax-M2.1 in Docs (#15538) Roger Young 2025-12-24 07:11:52 +08:00
• cf81737693 Add kv_transfer_total_mb to metrics (#15667) Lianmin Zheng 2025-12-23 11:16:55 -08:00
• aa6ac96635 [model-gateway] Fix tokenizer caching and improve error handling (#15695) Simo Lin 2025-12-23 10:33:38 -08:00
• 0d0367e9d0 Tiny fix CI (#15696) Liangsheng Yin 2025-12-24 01:51:08 +08:00
• bd572360f3 Tiny apply gsm8k mixin to ngram test (#15606) Liangsheng Yin 2025-12-24 01:30:26 +08:00
• 5f3a47d8a7 [model-gateway]: add gRPC router embeddings endpoint implementation (#15273) ratish 2025-12-23 21:12:22 +04:00
• 80ae2229d3 [model-gateway] Optimize router selection with lock-free snapshots (#15672) Praneth Paruchuri 2025-12-23 22:39:30 +05:30
• 705287b2e5 Tiny add more information in retract logging. (#15694) Liangsheng Yin 2025-12-24 00:43:36 +08:00
• 762846531f Fix Illegal Memory Access when fa3 + spec + topk + page_size > 1 (#15469) Yubo Wang 2025-12-23 08:13:57 -08:00
• dd620987d1 [model-gateway] Replace tokenizer with tokenizer registry for dynamic tokenizer loading in gRPC router (#12968) Arthur Cheng 2025-12-23 07:58:47 -08:00
• 53f974b973 fix: potential crash for missing stream attribute (#15644) shuwenn 2025-12-23 23:24:31 +08:00
• 6a5764a719 Super tiny add test_soft_watchdog to nightly (#15692) fzyzcjy 2025-12-23 22:46:15 +08:00
• 291f11ae39 [Minor] Enhance JIT kernel and add dev docs (#14570) DarkSharpness 2025-12-23 22:34:59 +08:00
• d7301c89ba [Feature] support fastsafetensors (#15091) Teng Ma 2025-12-23 22:33:56 +08:00
• 758b9067a0 [CI] fix UT assert error in test_tokenizer_manager.py (#15646) shuwenn 2025-12-23 22:24:33 +08:00
• ffc23ef877 [diffusion] feat: generalize layerwise offloader to flux1 (#15633) ryang 2025-12-23 22:22:40 +08:00
• b3f83cc1c5 Fix pipeline parallelism doc typos (#15688) Shangming Cai 2025-12-23 22:19:15 +08:00
• c15fa1c577 [2/N] Update doc of Pipeline Parallelism with case study (#15684) Shangming Cai 2025-12-23 20:56:15 +08:00
• fa2966983a Support PP for zmq_to_scheduler (#15312) Tianyu Guo 2025-12-23 17:07:55 +08:00
• f9dd90ac35 fix: increasing H200 test timeout (#15600) Douglas Yang 2025-12-23 01:00:37 -08:00
• 66902e0f1b [AMD] CI - Improve image discovery with remote registry fallback (#15463) Bingxu Chen 2025-12-23 15:04:58 +08:00
• e50f356f6f [AMD] CI - Detect the aiter version and rebuild if needed (#15460) yctseng0211 2025-12-23 14:42:48 +08:00
• ac42797cf7 [CI] Enable retry logic for flaky CI tests (#14983) Alison Shao 2025-12-22 22:23:42 -08:00
• 883747ced1 [CI] Migrate Attention Backend tests to test/registered/attention/ (#15563) Alison Shao 2025-12-22 22:17:52 -08:00
• 989d4b3012 [CI] Migrate nightly tests to test/registered/ (#15582) Alison Shao 2025-12-22 22:16:32 -08:00
• bc3ca30023 [PD] Support fake decode for PD disaggregation without prefill node (#14628) Baidu-AIAK 2025-12-23 12:43:33 +08:00
• 061f41affc MoE: Skip SiLU/GELU activation for masked experts (#15539) yuchengz816-bot 2025-12-22 22:08:31 -05:00
• 82f1d6157f [GLM-ASR] GLM-ASR Support (#15570) Yuxuan Zhang 2025-12-23 09:37:55 +08:00
• 5e1a495c65 Improve engine customization interface (#15635) Lianmin Zheng 2025-12-22 14:24:16 -08:00
• 34013d9d5a chore: bump sgl-kernel version to 0.3.20 (#15590) sglang-bot 2025-12-22 12:32:34 -08:00
• 7759716786 bugfix[schedule]: Refactor sort method and add related UT (#13576) SeanWeiSean 2025-12-23 03:13:57 +08:00
• 3c882db3ad Adjust wrong mtp meaning introduce by mimo (#15632) Liangsheng Yin 2025-12-23 02:06:46 +08:00
• b736a1525a Tiny add back missing router per attempt response metric (#15621) fzyzcjy 2025-12-23 01:59:10 +08:00
• 6984837d82 Fix router gRPC mode launch error caused by async loading (#15368) fzyzcjy 2025-12-23 01:39:01 +08:00
• 2142881b7e [model-gateway] return 503 when all workers are circuit-broken (#15611) Simo Lin 2025-12-22 04:06:09 -10:00
• d77f3fccbf [Diffusion] Support peak memory record in offline generate and serving (#15610) Xiaoyu Zhang 2025-12-22 21:21:21 +08:00
• 828dec1c7c [VLM] Tiny: Unify VLM environment variables (#15572) Yuan Luo 2025-12-22 17:32:37 +08:00
• 575a49dc0e [diffusion] chore: remove default post-denoising dit offload in local mode (#15573) ryang 2025-12-22 17:01:35 +08:00
• e62e17442e Tiny enable soft watchdog in CI for stuck without logs (#15616) fzyzcjy 2025-12-22 17:01:04 +08:00
• d5431ff894 Tiny add stuck simulation (#15613) fzyzcjy 2025-12-22 17:00:18 +08:00
• 454a2544f2 Support soft watchdog for tokenizer/detokenizer/dp-controller processes (#15607) fzyzcjy 2025-12-22 16:59:16 +08:00
• 89619a9904 Tiny avoid EnvField misuse (#15612) fzyzcjy 2025-12-22 16:20:47 +08:00
• cb30d056e3 add decode round robin policy (#15164) Hexq0210 2025-12-22 14:48:52 +08:00
• 677930c276 Add glm-4.6-fp8 with/without mtp in nightly ci (#15566) Qiaolin Yu 2025-12-21 22:27:02 -08:00
• beae3f961c Adapt fixture-kit to gsm8k mixin (#15599) Liangsheng Yin 2025-12-22 14:19:35 +08:00
• 1167867e8d [model-gateway] add retry support to OpenAI router chat endpoint (#15589) Simo Lin 2025-12-21 18:28:51 -10:00
• ad7f35fb9d Support gsp send routing id in bench serving (#15588) fzyzcjy 2025-12-22 11:32:24 +08:00
• f4100732b8 Tiny fix bench serving GSP mode cache file strategy (#15587) fzyzcjy 2025-12-22 11:27:55 +08:00
• 468931b572 [Tiny]Move deepseek fp4 cutlass moe test to per-commit test (#15565) Baizhou Zhang 2025-12-21 18:08:07 -08:00
• 796969caf0 Optimize Rust CI builds with proper sccache configuration (#15581) Simo Lin 2025-12-21 15:32:13 -10:00
• 122c250336 [model-gateway] add retry and circuit breaker support to gRPC routers (#15585) Simo Lin 2025-12-21 15:12:36 -10:00
• a3a552232d [feature] support hicache-3fs usrbio lib build for ubuntu24.04 (#15230) FlyPanda 2025-12-22 09:00:38 +08:00
• 0bf95e6d69 Fix type mismatch in LoRA batch validation causing assertion failures (#15427) Chenxi Li 2025-12-21 15:54:41 -08:00
• a92de891b8 Split dpsk fp4 4 gpu tests and move the mtp part to real stage b (#15553) Qiaolin Yu 2025-12-21 15:51:43 -08:00
• b9d786059f [model-gateway] refactor WorkerManager with fan_out helper and thin handlers (#15583) Simo Lin 2025-12-21 13:29:37 -10:00
• 1354063a48 [model-gateway] add WorkerService abstraction for worker business logic (#15580) Simo Lin 2025-12-21 12:35:35 -10:00
• 254de6d2fd Add triton_fused_moe config for GLM-4.6-FP8 tp8 blackwell (#15569) Qiaolin Yu 2025-12-21 13:41:22 -08:00
• 350fbbf4dc fix ds3.2 nsa backend prefill TBO (#14901) Jincong Chen 2025-12-22 05:16:46 +08:00
• a39126672a chore: bump sgl-kernel version to 0.3.20 (#15564) sglang-bot 2025-12-21 13:15:23 -08:00
• e1dcd0df3e [model-gateway] minor code clean up (#15578) Simo Lin 2025-12-21 11:00:32 -10:00
• 393e2f9b62 Revert "[FusedMoE] Fix fused w13 tp sharded weight loading" (#15579) Liangsheng Yin 2025-12-22 04:15:24 +08:00
• 8766a1dd24 [1 / N] Clean up logprob utils (#15509) Liangsheng Yin 2025-12-22 03:12:25 +08:00
• 1d9ba2ce4d [model-gateway] Use UUIDs for router-managed worker resources (#15540) shuwenn 2025-12-22 01:40:03 +08:00
• ef001fb811 [model-gateway] /parse/easoning and parse/function_call for sgl-model-gateway (#15568) Haoran Wang 2025-12-22 01:39:24 +08:00
• c69c1c4f07 [CI] Fix AMD CI to exclude multimodal_gen from main_package filter (#15558) sunxxuns 2025-12-21 03:58:17 -05:00
• bed301a5ac [Feature] Enable return routed experts (#12162) Junrong Lin 2025-12-21 15:16:43 +08:00
• 8fe3e37468 Support piecewise cuda graph for dsv3 fp4 (#15531) Ke Bao 2025-12-21 14:50:32 +08:00
• 6014365564 [EAGLE] Fix slow Triton compilation in EAGLE KV cache copy by chunking large num_locs_upper (#15111) YAMY 2025-12-20 22:14:14 -08:00
• fc05acc2c7 [FusedMoE] Fix fused w13 tp sharded weight loading (#15432) Yinghai Lu 2025-12-20 21:59:36 -08:00
• 1ed946680a [diffusion] bench: improve bench_serving by adding more controlling args (#15554) Mick 2025-12-21 13:37:45 +08:00
• d7fbe73bf2 [diffusion] chore: minor improvements and typo-fixing (#15556) Mick 2025-12-21 13:37:10 +08:00
• 42bff706df [diffusion] profiling: simplify --perf-dump-path JSON output (remove duplicate denoise steps) (#15537) Xiaoyu Zhang 2025-12-21 12:56:46 +08:00
• 26704c23c0 [model-gateway]: Tool parser for glm47 (#15520) Haoran Wang 2025-12-21 12:07:51 +08:00
• 43b7c17474 [CI] Fix /rerun-stage command by using requests for workflow dispatch (#15447) Alison Shao 2025-12-20 19:50:56 -08:00
• 96740d6983 feature: revamp nightly tests with combined runner (#15324) Douglas Yang 2025-12-20 19:25:22 -08:00
• 9a3bdf2c95 [CI] Migrate CUDA Graph tests to test/registered/cuda_graph/ (#15436) Alison Shao 2025-12-20 18:55:39 -08:00
• 4b351f6b95 Apply new moe align block size kernel (#14134) Xiaoyu Zhang 2025-12-21 10:13:32 +08:00
• 7fa4906f4f [sgl-kernel] Streamline kernel size report (Top 20 only) and clean up (#15552) Xiaoyu Zhang 2025-12-21 10:00:47 +08:00
• 050f108c29 Optimize Bailing-MoE with FlashInfer Fused All-Reduce (#15526) Yuan Luo 2025-12-21 09:34:04 +08:00
• 47cdb65a45 fix: update argument extraction in R1 chat template (#15547) Xinyuan Tong 2025-12-21 01:18:49 +00:00
• 1d90b194b2 [model-gateway] bugfix: backward compatibility for GET endpoints (#15413) shuwenn 2025-12-21 04:37:53 +08:00
• 537ef18d17 [model-gateway] Optimize WASM Runtime with Instance Pooling and Component Caching (#15515) Praneth Paruchuri 2025-12-21 01:57:02 +05:30
• 69412ccb1f [model-gateway] add model gateway multi-arch docker build, test and document docker image (#15544) Simo Lin 2025-12-20 09:56:59 -10:00
• d3885d4ba6 Fix docker gateway image name and add latest tag (#15542) Simo Lin 2025-12-20 09:05:03 -10:00
• 0a346d3bd9 feat: Add limit-mm-data-per-request argument to server arguments (#15418) Xinyuan Tong 2025-12-20 18:44:47 +00:00
• bc18cb8631 [diffusion] refactor: change zmq socket type to router for scheduler (#15479) Mick 2025-12-21 00:39:54 +08:00
• bee8ac5b88 [diffusion] doc: add --perf-dump-path section to profiling doc (#15533) Xiaoyu Zhang 2025-12-21 00:15:27 +08:00
• 41bd76e18b [diffusion] log: fix wrong use of suppress_other_loggers (#15534) Mick 2025-12-20 23:43:55 +08:00
• c6ca1b3afc [diffusion] chore: allow all attention backends if not specified (#15530) Mick 2025-12-20 23:14:55 +08:00
• 8999ce754f [diffusion] perf: support zero-cost weight offload and overlap with compute for wan-series (#15511) Xiaoyu Zhang 2025-12-20 22:52:40 +08:00
• dce2ed4467 fix MiMo-V2-Flash typo (#15536) Yingchun Lai 2025-12-20 22:49:06 +08:00
• 019517a356 [VLM] Support ViT Piecewise CUDA Graph for Qwen3-VL (#15320) Yuan Luo 2025-12-20 21:00:07 +08:00
• 1f1f05a85e vlm: refactor engine vlm params and support processor output as input (#14091) mlmz 2025-12-20 18:31:24 +08:00
• 165f5c04cb Optimize MiMo-V2-Flash by flashinfer fused allreduce (#15464) Yuan Luo 2025-12-20 17:45:18 +08:00
• 3e01f3a533 [NPU] [BUGFIX] [CRITICAL!] Fix NPU inference (torch_npu._npu_reshape_and_cache() crash) (#15484) Артем Савкин 2025-12-20 09:53:46 +03:00
• 51e2eaa458 [AMD] Support fast_topk kernels in sgl-kernel (#15172) Hubert Lu 2025-12-20 14:19:09 +08:00