Commit Graph

Select branches

Hide Pull Requests

main

patches

sm86

Mono Color

• fcaa84efa7 [BugFix] Gate MRV2 mixed sparse-MLA warmup on max_num_seqs > 1 (#47050) Nick Hillandziminghuang 2026-06-30 16:31:27 +01:00
• 9e84ec8648 [Refactor] Remove dead minimax allreduce rms kernel (#46842) Wentao Ye 2026-06-30 11:29:21 -04:00
• d8f483dc30 [Spec Decode] Fix hidden-state extraction block size for hybrid verifiers (#46301) 3 people 2026-06-30 18:19:51 +03:00
• dc148dc4d7 [CI][Bugfix] Fix Hybrid SSM NixlConnector PD prefix cache test (2 GPUs) (#47157) Nicolò Lucchesi 2026-06-30 17:14:13 +02:00
• 7cf7cbcd95 [Bugfix] MiniCPM-V 4.6: fix grid rows/cols swap in placeholder generation (#45918) tc-mb 2026-06-30 23:12:44 +08:00
• c231d1f290 fix(security): bound tokenizer work when explicit truncation_side is set (#47007) Juan Pérez de Algabaandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-30 17:08:51 +02:00
• db808b3961 [Model Runner V2][Spec Decode] Implement block verification for rejection sampling (#46781) Giancarlo Delfin 2026-06-30 10:07:24 -05:00
• 00ebf19cca [Bugfix][Quant] Raise actionable error instead of bare assert for group-size/TP mismatch (#46230) (#46236) Arsalan Shakil 2026-06-30 17:57:14 +03:00
• ded6676458 [Bugfix] Seed RayExecutorV2 TCPStore port by DP rank to avoid collisions (#45960) Seiji Eicherandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-30 07:37:34 -07:00
• 7a327f0b4f [Rust Frontend] Simplify unit tests with shared TestTokenizer (#47125) Bugen Zhao 2026-06-30 22:34:43 +08:00
• 1ab9522935 Remove more unnecessary load_weights methods (#47058) Harry Mellor 2026-06-30 15:22:16 +01:00
• 0fc2512094 [KV Offload] Pass ScheduleEndContext to on_schedule_end hook (#46450) Ronen SchafferandOr Ozeri 2026-06-30 17:07:12 +03:00
• 62c7d8009f Forward fix nightly errors from #44589 (#47151) Harry Mellor 2026-06-30 15:02:34 +01:00
• ab80b3dff4 [CI/Build] Bump PyNvVideoCodec version (#47139) Isotr0py 2026-06-30 21:38:46 +08:00
• 91055efd36 [XPU] C++ implementation for get_memory_info (#47134) Qiming Zhang 2026-06-30 06:34:47 -07:00
• 3675bcff67 [Rust Frontend] Refactor TLS serve path with unified MaybeTlsListener (#47101) Bugen Zhao 2026-06-30 21:31:58 +08:00
• bdbd7278b6 [Rust Frontend] Extend renderer/parser roundtrip tests to support token ids (#47110) Bugen Zhao 2026-06-30 21:27:45 +08:00
• 5dc36a4fa5 [Model] Remove Tarsier, Tarsier2 (#47143) Harry Mellor 2026-06-30 14:20:33 +01:00
• aab7af0bcb [Bugfix][ROCm][MLA] Pass q/kv dtypes to get_mla_metadata_v1 in FP8 decode (#46997) peizhang56andTJian 2026-06-30 05:31:16 -07:00
• 536047755e Bump actions/checkout from 6.0.1 to 7.0.0 (#33057) dependabot[bot]anddependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> 2026-06-30 13:16:20 +01:00
• 1907d3854a [Bugfix] Reject negative values for max_logprobs and long_prefill_token_threshold (#44002) 4 people 2026-06-30 20:01:03 +08:00
• ea9ddf59fc [XPU][CI] Enable shared loader test (#45977) Chaojun Zhang 2026-06-30 19:20:33 +08:00
• 8cf7c4d8ad [Attention Backend] add HPC-Ops Attention backend (#46020) 3 people 2026-06-30 18:17:43 +08:00
• 8e9d70fdd5 [Kernel][XPU] Adjust kernel unit tests for XPU (#45140) Agata DobrzyniewiczandKunshang Ji 2026-06-30 11:57:27 +02:00
• 364ee36af1 fix(security): prevent image decompression bomb OOM denial of service (#47010) Juan Pérez de Algaba 2026-06-30 11:39:22 +02:00
• 06fae69114 [Misc] Mistral label alert (#47132) Nicolò Lucchesi 2026-06-30 11:02:07 +02:00
• 14f8660a18 [CI/Build] Add CPU test dependency pre-commit hooks (#47032) Li, JiangandHarry Mellor 2026-06-30 15:59:13 +08:00
• aed541def4 [Bugfix][Responses] Set completed status for Harmony function calls (#46945) 3 people 2026-06-30 13:25:14 +05:30
• 2bc20e8aba [Frontend] Add Streaming Parser Engine and new Kimi k2.5/k2.6/k2.7 Parser (#46610) Chaunceyandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-30 15:53:17 +08:00
• 8cc242335d [XPU] Optimize XPU worker shutdown logic to prevent resource leak (#46433) Chaojun Zhang 2026-06-30 15:27:21 +08:00
• ba22cb6765 [ROCm][Ray][CI] Keep assigned GPU visible for weight transfer (#47000) Andreas Karatzas 2026-06-30 01:59:18 -05:00
• 81bcced482 [Bugfix][ROCm] Preserve MoE weight padding for unquantized Triton path (#46381) Uros Markovic 2026-06-30 08:47:57 +02:00
• fb42e5219e [Platform] Replace torch.cuda.mem_get_info with torch.accelerator.get_memory_info (#44825) Kunshang Ji 2026-06-30 14:39:52 +08:00
• 0feca7ffa8 PD disagg with Mooncake Connector: GDN support (Qwen3.5) and MLA support (Deepseek-V4-Flash) (#46807) Dakai An 2026-06-30 14:29:04 +08:00
• 97b5ce5c39 [Bugfix] Raise VLLMValidationError for non-integer logit_bias keys (#46612) Muhammad FawazandMahad Durrani 2026-06-30 11:18:59 +05:00
• 4236514098 [ROCm][CI][Multimodal] Use ROCm-aware FA availability check for Unlimited-OCR (#47004) Andreas Karatzas 2026-06-30 01:03:13 -05:00
• e45c8a9f4b [Rust Frontend] Start current wave for a stale DP FirstRequest (#46833) Blas Rodriguez Irizar 2026-06-30 06:13:09 +01:00
• b153dd3f28 [Bugfix] Use larger workspace size for Flashinfer MLA LSE (#47074) Wei Zhao 2026-06-30 01:11:03 -04:00
• 930f8dc0a1 [Bugfix][Rust Frontend] Reject prompt_logprobs for streaming generate (#46839) Reid 2026-06-30 13:10:07 +08:00
• a16dbd5b85 [Rust Frontend] Avoid LoRA registry scans without active LoRA requests (#47040) Reid 2026-06-30 12:58:19 +08:00
• bec232a914 Secondary tier implementation for PD disaggregation (#42285) 3 people 2026-06-30 07:51:44 +03:00
• b5c9e1ac33 [LoRA] Add language-backbone LoRA support for MiniCPM-V 4.6 (#46740) linitra24andJee Jee Li 2026-06-30 12:19:31 +08:00
• ae2c4f3db7 [XPU][UT]Fix xpu pass_config.fuse_norm_quant assert issue (#46804) Yejing LaiandKunshang Ji 2026-06-30 12:13:44 +08:00
• fca432e60a [Bugfix] Propagate default stop_token_ids to per-request SamplingParams (#35076) ganesh 2026-06-29 23:10:09 -05:00
• af1ee8c475 fix(config): reject negative max_logprobs (except -1) and long_prefill_token_threshold (#44070) hclandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-06-30 12:02:36 +08:00
• 5b4cb69523 [Bugfix][MLA] Fix LSE log-base mismatch in DCP + FlashInfer MLA decode (#47079) Summer YangandClaude 2026-06-29 19:15:02 -07:00
• 9fc0c08026 [ROCm][CI] Make tests/v1/shutdown an importable package (#47085) peizhang56andClaude 2026-06-29 19:01:27 -07:00
• f2b5fabb23 [ROCm][CI] Move LM Eval Large Models (8 GPUs) to mi300 pool (#47094) 3 people 2026-06-29 18:59:08 -07:00
• b8cb75b149 [Rust Frontend] Add static HTTPS and mTLS support for HTTP and gRPC (#45890) 3 people 2026-06-30 07:45:59 +06:00
• 43916891b2 [GDN] Improve kkt kernel of CuteDSL prefill backend (#46346) Thien Tran 2026-06-30 09:34:18 +08:00
• cda05ee8c4 [Bugfix][Reasoning] Fix thinking_token_budget not enforced on re-entry after forced end (#43757) 3 people 2026-06-29 18:04:25 -07:00
• 77654d080c [KVTransfer] MultiConnector: merge kv_transfer_params dicts across connectors (#46777) weishu 2026-06-29 17:25:05 -07:00
• 75698e60b3 [Bug] Fix sparse attention issue for GLM5.2 non-torch compile path (#47083) Wentao Ye 2026-06-29 18:45:53 -04:00
• 8632c884dc [ROCm][CI] Use spawn around the threaded OTLP test (#47003) Andreas Karatzas 2026-06-29 16:34:05 -05:00
• c3734e8334 [CI][Bugfix] Add cohere_melody to ROCm test requirements (#47072) peizhang56andClaude 2026-06-29 14:29:47 -07:00
• 53f7553f09 [ROCm][DeepEP] Stabilize high-throughput DBO for DP+EP (#46990) Andreas KaratzasandTyler Michael Smith 2026-06-29 16:28:02 -05:00
• 4eb227992a [ROCm][CI] Make memory sampling less racy in tests and sleep mode (#45490) Andreas KaratzasandCodex 2026-06-29 16:26:41 -05:00
• ebcf511ec3 [ROCm][CI] Soft Fail Spec Decode Ngram + Suffix and Entrypoints Integration (LLM) AMD Mirrors (#47067) Micah Williamson 2026-06-29 16:24:08 -05:00
• 8fc1b2d046 Fix FA4 dynamic_causal for full attention layers (#46659) Matthew Bonanni 2026-06-29 17:23:34 -04:00
• 5316638a5e Fix transient dependency issues caused by requirements/common.txt (#47015) Harry Mellor 2026-06-29 22:20:33 +01:00
• 61ab70ec3b [Model Runner V2] support mamba hybrid models align prefix cache (#42406) zhrrr 2026-06-30 05:09:16 +08:00
• a309d4fe60 Support DCP with FlashInfer MLA (#43729) Woosuk Kwon 2026-06-29 13:24:29 -07:00
• 72f639927f [XPU] [RMSNorm] revert weightless change on xpu (#46987) zofiaandKunshang Ji 2026-06-30 03:03:06 +08:00
• 8ad4a01825 [ModelRunner V2] Simplify recent UnlimitedOCR-related changes (#46975) Nick Hill 2026-06-29 17:56:17 +01:00
• 7be582697b [Bugfix] Fix DeepseekV2Model hidden_size (#46986) Jee Jee Li 2026-06-30 00:44:05 +08:00
• 030c9523bd [Perf][1/N] Expand Triton kernel warmup coverage, DSv4 (#46634) Roberto L. CastroandLucas Wilkinson 2026-06-29 18:40:34 +02:00
• 4708292d48 Bump flashinfer version to 0.6.13 (#46683) Wei ZhaoandJee Jee Li 2026-06-29 12:30:57 -04:00
• debec6440b Add MiniMax-M3 modelopt nvfp4 support (#46756) Jason LiandXin Li 2026-06-29 12:29:39 -04:00
• c8fb2963bd [FS-Offloading] Batch Lookup in C (#46713) Varun Sundar RabindranathandVarun Sundar Rabindranath 2026-06-29 12:28:32 -04:00
• 379acd4e4f [Bugfix][Quantization] Fix W8A8 int-quantized scheme selection regression (#46860) HDCharles 2026-06-29 11:55:42 -04:00
• 07d33e575b [MyPy] Fix mypy incompatible assignment errors in LRUCacheLoRAModelManager (#44657) Martin Hickey 2026-06-29 16:42:35 +01:00
• 36bbecd643 [BugFix] Revert "[KV Offload] Use background thread for mmap / cpu_tensors pinning" (#46958) Varun Sundar RabindranathandVarun Sundar Rabindranath 2026-06-29 10:54:34 -04:00
• 6149187a4c [Kernel] Triton MLA logits workspace (#46819) Nicolò Lucchesi 2026-06-29 16:54:29 +02:00
• 49e28e8e91 [Kernel][Helion][1/N] Add Helion kernel for fused_qk_norm_rope (#44010) Xiaohong (Sean) Chen 2026-06-29 10:54:15 -04:00
• 0ca39c4f1f [Bugfix] Capture final-layer aux hidden state in deepseek_v2 backbone (#46973) Michael GoinandClaude Opus 4.8 2026-06-29 10:00:31 -04:00
• 6185d73882 [Rust Frontend] Keep literal "null" string for string-typed tool params (#46827) Blas Rodriguez Irizar 2026-06-29 14:46:33 +01:00
• bc8481af09 [MoE Refactor] Standardize Humming MoE experts + utilities (#43373) bnellnmandRobert Shaw 2026-06-29 09:19:29 -04:00
• 59575da46d [XPU] exclude unsupported models for test_tensor_sechma.py (#47008) Yan MaandKunshang Ji 2026-06-29 20:30:28 +08:00
• 3483240b7e [Frontend] Consolidate scale out entrypoints (#44512) wang.yuqi 2026-06-29 18:18:53 +08:00
• eddfd4cf21 [Perf][2/N] Expand Triton kernel warmup coverage, Qwen (#46750) Roberto L. Castro 2026-06-29 12:10:07 +02:00
• a4e3cb40d0 [mypy] Enable mypy for tests directory (#47018) Martin Hickey 2026-06-29 10:29:09 +01:00
• ab132ee98b Fix model info cache for package models (#46567) soaringk 2026-06-29 17:17:54 +08:00
• e186107870 [Bugfix] Use native SiLU activation in CPU fused MoE (#45961) Alden LoboandAlden Lobo 2026-06-29 04:12:20 -05:00
• 0e207dac78 [Bugfix] Transformers backend: apply learned lm_head.bias for tied-embedding models (#46835) JohnandHarry Mellor 2026-06-29 04:59:15 -04:00
• 9e86352c60 [CI Failure] Add transformers version check for openai/privacy-filter (#47011) wang.yuqi 2026-06-29 16:57:26 +08:00
• 5051698e41 Remove unnecessary load_weights methods (#44589) Harry Mellor 2026-06-29 09:52:23 +01:00
• db28ae2d07 [ROCm][CI] Explicitly tear down multimodal offline LLMs (#46999) Andreas Karatzas 2026-06-29 02:59:24 -05:00
• f6bb8682ee Fix docs on main (#47009) Harry Mellor 2026-06-29 08:50:57 +01:00
• 4559c43a95 [MM][CG] Gemma3 Encoder CUDA Graph (#43591) SoyaazzandIsotr0py 2026-06-29 12:52:00 +08:00
• 5274c1181d [Rust Frontend] Add Harmony Renderer for GPT-OSS (#46800) Bugen Zhao 2026-06-29 11:39:04 +08:00
• 58d6a6e60a [CPU] Support cpu compressed-tensor w8a8 int8 moe (#42920) Yuwen Zhou 2026-06-29 11:04:05 +08:00
• a2abce646f [EPLB] Mask padding in EPLB load recording (#38128) Ilya MarkovandMarkov Ilya 2026-06-29 04:43:58 +02:00
• 311ad689ad Remove boilerplate missed by #46820 (#46956) Harry Mellor 2026-06-29 01:11:17 +01:00
• 0472436541 [Spec Decode] Avoid redundant hidden-states gather in draft prefill (#46968) Woosuk Kwon 2026-06-28 17:04:01 -07:00
• 4dfbf1503b [Model] Add support for openai/privacy-filter (#41026) 3 people 2026-06-29 01:18:22 +02:00
• 95528527ea [Bugfix][Mooncake] Fix Mooncake lookup prefixes with DCP > 1 (#46855) Wei Zhao 2026-06-28 17:36:23 -04:00
• c2127a25c7 [ROCm][CI] Fix rlhf_async_new_apis Example On ROCm (#46895) 3 people 2026-06-28 12:50:30 -05:00
• 03c6d01c30 [OCP MX ] Add back emulation to available OCP MX backends list (#46629) fxmarty-amdandAndreas Karatzas 2026-06-28 19:43:19 +02:00
• 4b643c463e [GLM5] Fix minor typo (#46961) Woosuk Kwon 2026-06-28 08:37:00 -07:00
• 7544286b04 [Bugfix] Transformers backend: recompute mm_token_type_ids per request for M-RoPE (#46552) Gonzague de CarpentierandHarry Mellor 2026-06-28 17:19:28 +02:00