Commit Graph

Select branches

Hide Pull Requests

main

patches

sm86

Mono Color

• d4b4562917 [XPU] Bump vllm_xpu_kernels to v0.1.11.1 (#48942) Artur FierkaandClaude 2026-07-17 14:43:23 +02:00
• 7b3192523e [Bugfix]Fix transformer backend failed: AttributeError: 'Parameter' object has no attribute 'weight_loader' (#48699) 3 people 2026-07-17 19:43:44 +08:00
• 4c6e2e4b30 [XPU][UT]fix _POSSIBLE_KERNELS error on XPU (#47516) Yejing Lai 2026-07-17 19:19:05 +08:00
• 8502958810 [XPU] support HND layout (#47975) liuzhenwei 2026-07-17 18:54:34 +08:00
• ce4bdcbda4 [Bugfix] Enable FlashAttention MLA prefill for Mistral Small 4 head dims (#48855) Julien DenizeandNicolò Lucchesi 2026-07-17 12:07:00 +02:00
• d5b1ec2684 [XPU] allow forcing flash attn for mm_prefix (#48828) liuzhenwei 2026-07-17 17:44:18 +08:00
• 867ff69733 [CI] Gate non-default release wheel builds (#48772) Kevin H. LuuandOpenAI Codex 2026-07-17 02:16:44 -07:00
• 109b736b86 [docs] preserve page path in stable-docs announcement link (#48839) Sage 2026-07-17 11:56:45 +03:00
• 69d4f5ef63 [Bugfix][Multimodal] Fix Qwen3-Omni use_audio_in_video with mixed image/video inputs (#46213) wendadawenandTianyu Guo 2026-07-17 16:31:16 +08:00
• 426d48bfa1 [KV Offload] Add optional tier locality to FS/OBJ KV events (#48281) 3 people 2026-07-17 00:31:52 -07:00
• 26c909ed74 [Model] Support TranslateGemma-12b-it (#41599) 4 people 2026-07-17 15:17:59 +08:00
• fb1d8ccaf5 [rl] Stateful Trainer Send: New Abstractions [1/N] (#48042) Aaron HaoandSumanth R Hegde 2026-07-17 00:11:06 -07:00
• 9354f22204 [Rust][Benchmark] Port in vllm-bench (#48107) Roy WangandBugen Zhao 2026-07-17 14:25:29 +08:00
• 17fdd42100 [Bugfix][Attention] Preserve post-load tensors across weight reloads (#48251) aoshen02 2026-07-17 14:15:26 +08:00
• 472d330c21 Add blocks_per_chunk configuration for KV offloading to support heterogeneous KV cache groups (#48878) Debasish MohantyandOr Ozeri 2026-07-17 11:30:12 +05:30
• 3b6c96a101 [Bugfix][Pooling] Fix wrong scores for chunked prefill under torch.compile (#48901) 3 people 2026-07-16 22:19:12 -07:00
• 4d4e04f452 [Render] Add round trip parity test and docs for derender (#48617) Martin Hickey 2026-07-17 06:02:46 +01:00
• 67fe73b2b4 [CI] Extend max-model-len for test_parsable_context to allow reasoning to finish (#48873) Micah Williamson 2026-07-16 22:36:44 -05:00
• ee8f36d0b3 [Warmup] Show CuTeDSL compilation progress (#48881) 10 people 2026-07-16 20:17:51 -07:00
• f3e9497e92 [Model] Add Inkling LoRA support [4/N] (#48884) 10 people 2026-07-16 18:52:15 -07:00
• fe784ff22e [M3] Improve indexer for long-context decode (sm100) (#48582) Thien Tran 2026-07-17 09:12:15 +08:00
• b88abb5036 [Misc] Remove orphaned env vars and stale env-var references (#44749) Daoyuan Li 2026-07-16 17:00:47 -07:00
• 67f9046e4a [Bugfix] Sparse MLA: enable fp8_ds_mla dense prefill (#48642) 3 people 2026-07-16 18:44:04 -04:00
• f17be06fbe [Perf] Optimize clamp to clamp_ (#48143) Wentao Yeandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-16 18:41:07 -04:00
• 2cab53ddee [Model][Hardware][AMD]: Part 1/2 -> Enable e2e QK Norm + RoPE + KV Cache runtime fusion for Qwen3-30B-A3B on ROCM_AITER_FA, and ROCM_AITER_UNIFIED_ATTN (#42749) Jack HuandDouglas Lehr 2026-07-16 18:39:04 -04:00
• ab0a20d151 [Docs] Add Phi-3.5-mini-instruct to batch invariance tested models (#46396) 3 people 2026-07-17 01:05:58 +03:00
• 4a394bfcda [Spec Decode][DSpark] Add Gemma4-12B DSpark draft model (#47216) Hangrui CaoandDiegoCao 2026-07-16 14:51:47 -07:00
• c95c663049 [Quant] Add nvfp4_per_token online MoE quantization (#48538) Michael Goin 2026-07-16 17:25:27 -04:00
• ab3c1aedf3 [Bugfix] Fix activation quantization dispatch for WNA4Int/WNA8Int (#48785) HDCharles 2026-07-16 17:13:02 -04:00
• fb5ec0dc9e [Model] Add Inkling MTP=1 support [3/N] (#48869) 10 people 2026-07-16 13:27:21 -07:00
• 971dac2caa [Bugfix][KV-transfer] MoRIIO: retry RDMA send-queue-full backpressure instead of failing the read (#47495) 3 people 2026-07-16 13:02:27 -07:00
• efa2e424f6 [Helion] Fix degenerate scale_ub in kernel input generators (#48868) Shangdi Yu 2026-07-16 12:57:45 -07:00
• 02bf9c7907 Fix Quark mxfp4 quantized model loading issue under mtp (#46757) xiao-llmandAndreas Karatzas 2026-07-16 14:58:47 -04:00
• f61163e6c7 [Model] Add Hopper FA4 relative attention for Inkling (#48858) Woosuk Kwon 2026-07-16 11:26:52 -07:00
• 626c90b2d5 [Refactor] Move fla to third party (#48500) Wentao Ye 2026-07-16 14:22:36 -04:00
• 251f7e478e [Model] Add PW CUDA graph support for Inkling [2/N] (#48822) 10 people 2026-07-16 10:28:56 -07:00
• ce65385618 [KV Offload] Split tiering_lookup_delay into sync/async histograms (#47679) Srinivas Krovvidiandsrinivas_oo7 2026-07-16 12:03:45 -05:00
• 7d56fe2adc [ROCm][CI] Avoid HIP init at config time via lazy aiter import in Quark OCP-MX (#48015) music-dino 2026-07-16 18:47:08 +02:00
• 75bdad40b5 [Bug][Quantization] Fix humming is_layer_skipped for compressed-tensors "re:" ignore entries (#48507) Zhongdongming Dai 2026-07-16 07:47:42 -07:00
• d08eebad16 [Perf][MoE] Write FlashInfer combine into final output (#47156) Samuel NordmannandCodex 2026-07-16 16:21:03 +02:00
• 3e90d015ba [Frontend] Overlap preprocessing and computation for pooling models offline inference (#47699) wang.yuqi 2026-07-16 22:00:20 +08:00
• 7cd1d57b74 [CI/Build][Docker] Bump nvidia-cutlass-dsl to 4.6.0 and drop packaging workarounds (#47442) Artem PerevedentsevandMatthew Bonanni 2026-07-16 16:51:41 +03:00
• b8168e33e0 [ROCm][Perf][DSV4] Enable split sparse decode on gfx942 (#46275) Tuukka Sarviandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-16 15:28:20 +03:00
• d803b44dbe [NIXL] Bump nixl to 1.3.1 (#47559) ovidiusm 2026-07-16 14:24:33 +02:00
• 530852f959 [KV Connector] Fix PD async scheduling race condition for hybrid attn models (#48481) 4 people 2026-07-16 13:42:37 +03:00
• a317bc5739 [Misc][Nixl] Unify _logical_to_remote_kernel_block_ids (#48717) Nicolò Lucchesi 2026-07-16 12:36:25 +02:00
• a9531edfa6 [KV Offload] Define clean backend configuration boundary (#48150) 5 people 2026-07-16 03:27:05 -07:00
• 8c3393f373 [Bugfix][Rust Frontend] Limit chat top_logprobs in responses (#48134) Reid 2026-07-16 17:41:44 +08:00
• 9f8cbfd8eb Vectorize prep xfer list creation (#48209) Ilia Yastrebov 2026-07-16 11:39:05 +02:00
• ea1d65fe6d [Rust Frontend] Add Seed-OSS tool parser (#47741) Chao-Ju Chenandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-16 17:28:02 +08:00
• f44f3d6f79 [Rust Frontend] Wait for mock engine endpoints before ZMQ connect (#47965) ReidandBugen Zhao 2026-07-16 17:20:31 +08:00
• cc706b05a5 [Bugfix][Rust Frontend] Detokenizer: avoid leaking prompt on zero-generated-token completions (#47707) JinYan SuandBugen Zhao 2026-07-16 17:11:21 +08:00
• 85e296950c BF16x3 router GEMM (#47973) Thien Tran 2026-07-16 17:04:20 +08:00
• dc9f845ddc [Rust Frontend] Fix mock engine test shutdown race (#48738) Reid 2026-07-16 15:27:43 +08:00
• 12f2c515a7 [Bugfix] Fix offloading set_ overflow for packed non-uniform KV caches (#48530) Elvir Crnčević 2026-07-16 09:04:39 +02:00
• 6570c9800c [Model] Add Inkling model support [1/N] (#48799) 10 people 2026-07-15 23:40:07 -07:00
• 8bfd683901 [Spec Decode] Add kv_cache_dtype to speculative_config to control separately from target (#48787) Michael GoinandClaude Opus 4.8 2026-07-16 01:49:23 -04:00
• 7dc2698632 [ROCm][CI] Set "highest" matmul precision for reference hf_runner in test_bert_for_masked_lm (#48784) Micah Williamson 2026-07-16 00:49:15 -05:00
• 59b964f37d fix(lora): validate LoRA rank is positive in PEFTHelper (#48437) ErenAta16 2026-07-16 08:22:52 +03:00
• 6a9f24aa8c [ROCm][CI] Fix cuda graph mem profile issue (#48764) Charlie FuandAndreas Karatzas 2026-07-15 23:23:18 -05:00
• ba47bb5be1 Bump flashinfer version to 0.6.14 (#47669) 5 people 2026-07-15 21:00:40 -07:00
• df8a0900df [BugFix] Don't apply weight in batch-invariant RMSNorm when has_weight=False (#48741) Asaf GardinandMichael Gokhman 2026-07-16 06:59:03 +03:00
• 2db39c7049 [Bugfix][Spec Decode] Fix eagle3 first-layer qkv_proj prefix for quantized drafts (#48068) qizixiandClaude Opus 4.8 2026-07-15 20:34:41 -07:00
• 3935829f89 [Docs] fix error key name (#48802) rongfu.leng 2026-07-16 11:16:16 +08:00
• 7746961277 [CI] Fix flaky lora test (#47375) qli88 2026-07-15 21:23:03 -05:00
• 5de1add806 [feature]Add int4 quantization support for emulation moe backend (#48451) qli88 2026-07-15 21:00:12 -05:00
• 915dffaa5f [Attention] Mirror Triton KV dtype checks in MLA (#47060) Mike G 2026-07-15 18:54:52 -07:00
• 81e13a0591 [Compilation] Skip x.size(dim) in _decompose_size_nodes (#42543) nemanjaudovic 2026-07-16 02:59:14 +02:00
• f95e3f0edb [Tests] Gate Step3VL under Transformers v5 (#44349) BRIJ RAJ KISHORE 2026-07-16 06:29:10 +05:30
• 5a65ba5f17 [Refactor] Move iteration logging to the frontend (#46647) 3 people 2026-07-15 20:59:05 -04:00
• 9d1c695be5 [XPU] Add DSpark speculative decoding support for DeepSeek-V4 (#47677) 3 people 2026-07-16 08:59:02 +08:00
• 3c1bc1fc0d [ROCm][Perf] Optimize sparse attention prefill kernel for DeepSeek-V4 (#48519) kliuaeandTJian 2026-07-16 08:58:59 +08:00
• 3a5e88e629 [Bugfix] Fix local speculators with dots in the name from classifying as custom_class (#48754) Michael Goin 2026-07-15 20:58:06 -04:00
• 0becb7486b [BugFix][MLA] Support kv_cache_dtype_skip_layers for MLA attention (#47309) 4 people 2026-07-16 08:06:11 +08:00
• 2dab187f75 [Perf] Optimize fused_topk_bias for DSv4, 1.5~2x kernel performance improvement (#47463) Wentao Ye 2026-07-15 19:40:55 -04:00
• 015b0320de Add giuseppegrossi to rocm label auto cc action (#48643) Giuseppe Grossi 2026-07-15 19:38:56 -04:00
• 4238b011a7 [Feature] Migrate moe sp support to non-torch compiled path for GLM5.2 (#47881) Wentao Yeandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-15 19:33:15 -04:00
• eb33ff34dd [ROCm][Perf] DSv4 two-stage compressor kernel for HCA prefill (#47718) kliuae 2026-07-16 07:31:51 +08:00
• 2bd8957627 [Bugfix][NVFP4 MoE] Pad gated intermediate to 64 for FlashInfer TRT-LLM shuffle (M%128) (#46880) Mike G 2026-07-15 14:37:15 -07:00
• 3034c8d389 [CI][PD] Add optional/nightly DSv4 Disaggregated eval (#42310) Nicolò Lucchesi 2026-07-15 23:04:54 +02:00
• ecf4aa5ce2 [Bugfix] Fix FlashInfer non-causal draft attention (DFlash/DSpark) on Blackwell (#48167) Michael GoinandClaude Fable 5 2026-07-15 15:44:01 -04:00
• 49e777cf08 [CI][ROCm] Retry failed Docker build steps once (#48773) Micah WilliamsonandAndreas Karatzas 2026-07-15 14:31:23 -05:00
• b7950e798f [Bugfix] Initialize draft CUDA-graph keys for the native draft_model proposer (#47460) avalliappan-nvidiaandClaude 2026-07-15 12:09:00 -07:00
• de100ffb62 [Docs] Document pooling config resolution (#48497) Taneem Ibrahimandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-15 14:24:16 -04:00
• 43cd340247 [Fix] Align OpenAI vllm_xargs value types across request schemas (#48252) Sage 2026-07-15 20:48:24 +03:00
• 1d99f0f421 [ROCm][BugFix] Triton W4A16 handling for GPTQ/AutoGPTQ qzeros layout (#47770) Giuseppe Grossiandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-15 12:55:47 -04:00
• 0885b51981 [CI][ROCm] Stabilize ci_base hash calculation and image handoff (#48746) Andreas Karatzas 2026-07-15 10:56:56 -05:00
• 6036bf110a [Kernel][Helion] Add Helion kernel benchmark script (#48512) Xiaohong (Sean) Chen 2026-07-15 11:43:06 -04:00
• 2fa63e0fff [Kernel][Helion] Helion kernel lazy registration (#48264) Xiaohong (Sean) Chen 2026-07-15 11:42:46 -04:00
• 61141ed265 [Hardware][XPU] Register batch-invariant kernels for XPU (#41934) 3 people 2026-07-15 17:19:44 +02:00
• 05eed72aec [ROCm] Re-enable cudagraph memory profiling, captured on the current stream (#48526) 3 people 2026-07-15 08:03:48 -07:00
• 5810e884f1 [Model] Add RobertaForTokenClassification / XLMRobertaForTokenClassification (#47991) Gopala-Krishna Char 2026-07-15 16:30:33 +02:00
• 615834ee58 [KVOffload][P2P] Well-known default host/port env vars and per-DP-rank control port (#47636) liranschourandmergify[bot] <37929162+mergify[bot]@users.noreply.github.com> 2026-07-15 15:22:56 +03:00
• 5811ed6a05 [Test][kv_offload] Fix flaky drain() helper in test_fs_tier.py (#48545) Chaojun Zhang 2026-07-15 19:53:47 +08:00
• 1b30ae4ca4 [Rust Frontend] Fix flaky tls_handshake_timeout_drops_silent_client test (#47873) Tahsin Tunan 2026-07-15 17:05:38 +06:00
• 4e04bcbce6 [Rust Frontend] Tolerate whitespace before the outer brace in JSON tool-call parsers (#48034) Tahsin Tunan 2026-07-15 17:03:37 +06:00
• 66b6c684ab [PD][Bugfix] Fix validation of cache shape for attn backends enforcing different kernel_block_size (#48125) Nicolò Lucchesi 2026-07-15 12:26:02 +02:00
• c0302d9497 [Bugfix] Fix parallel_tool_calls=null crash in Responses API from_request() (#48098) 4 people 2026-07-15 15:01:17 +05:00
• 313fae3e89 [Bugfix] Fix GLM5 config (#48711) Jee Jee Li 2026-07-15 17:55:39 +08:00
• 7aab6e2684 [ROCm][Bugfix] Enable the fp32 head_dtype torch.mm fast path on ROCm (#48688) Turner JabbourandClaude 2026-07-15 02:18:22 -06:00