3 people 3f55ad2f07 [Qwen3.8-Flash-Next] Fuse PLE residual and QSA output gate (#55309) ...

Signed-off-by: Canlin <canlinguosdu@gmail.com>
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: Kimi Code <noreply@moonshot.cn>

2026-09-14 12:16:25 +00:00

..

assets

[Bugfix] Align OpenCV video metadata timeline (#47099)

2026-06-30 20:43:42 +00:00

benchmarks

[Bugfix][Benchmark] Make streaming TTFT/E2E latency accounting consistent across endpoints (#55508)

2026-09-13 05:20:54 +00:00

compilation

[3/N] HiSparse: host-resident sparse-MLA decode hot-buffering (#53781)

2026-09-12 00:36:47 -04:00

config

[Performance] Add Triton kernel for Gemma3n sparse GELU (#48498)

2026-09-14 07:42:15 +00:00

cute_utils

[5/N][warmup][DSv4] Migrate NVIDIA CuTeDSL attention kernels (#53566)

2026-09-12 00:05:59 -04:00

device_allocator

[Core] Release NCCL communicator memory in sleep mode (#51485)

2026-09-01 14:31:16 +08:00

distributed

[Bugfix][NixlPush] Guard _remote_agents read in _do_send_reg_notif (X1) (#56317)

2026-09-14 18:37:12 +08:00

engine

[Perf][Kernel] Integrate DeepSelect TopK for the DSA sparse indexer (#56464)

2026-09-13 00:44:37 -07:00

entrypoints

[Frontend] Replace VLLM_ENABLE_SCALE_OUT_ENDPOINTS with --enable-scale-out (#55176)

2026-09-14 10:02:14 +00:00

inputs

[Feat][MM Hashing] include media_io_kwargs in multi-modal hashes (#54241)

2026-09-01 13:17:01 +00:00

ir

[Performance] Add Triton kernel for Gemma3n sparse GELU (#48498)

2026-09-14 07:42:15 +00:00

kernels

[Performance] Add Triton kernel for Gemma3n sparse GELU (#48498)

2026-09-14 07:42:15 +00:00

logging_utils

…

lora

[LoRA] Add LoRA support for DeepSeek-V4 Flash Vision (#55897)

2026-09-13 07:10:54 -07:00

model_executor

[ROCm][Perf] W4A16: keep skinny GEMM zero-points packed 4-bit (#54965)

2026-09-14 17:40:25 +08:00

models

[Qwen3.8-Flash-Next] Fuse PLE residual and QSA output gate (#55309)

2026-09-14 12:16:25 +00:00

multimodal

[Security] Cap Qwen-VL video sampling knobs (#56729)

2026-09-14 10:22:18 +00:00

parser

[Frontend] create unified Cohere parser (#56392)

2026-09-12 04:04:46 -04:00

platforms

[Performance] Add Triton kernel for Gemma3n sparse GELU (#48498)

2026-09-14 07:42:15 +00:00

plugins

[MISC] Cleanup deprecated parameters (#53559)

2026-08-24 10:19:04 -07:00

profiler

[CPU][Profiler] Group torch profiler tables by input shape when record_shapes is on (#56016)

2026-09-14 17:16:48 +08:00

ray

Propagate Flash Attention cache configuration to Ray workers (#49177)

2026-07-21 07:47:53 +00:00

reasoning

[Frontend] create unified Cohere parser (#56392)

2026-09-12 04:04:46 -04:00

renderers

[Bugfix][EPD] Preserve explicit multimodal UUIDs with caches disabled (#56432)

2026-09-13 10:08:30 +00:00

third_party

[XPU] Add forward_xpu to Mixer2RMSNormGated and FusedRMSNormGated (#54968)

2026-09-11 07:58:28 +08:00

tilelang_utils

[ROCm] Defer tilelang import through its import from vllm.tilelang_utils import tilelang and relaxed has_tilelang (#51159)

2026-08-13 13:12:40 -05:00

tokenizers

[Bugfix][Frontend] Support Responses text types in DeepSeek V4.1 (#56299)

2026-09-14 10:43:09 +08:00

tool_parsers

[Frontend] create unified Cohere parser (#56392)

2026-09-12 04:04:46 -04:00

tracing

…

transformers_utils

[Refactor] Normalize DeepSeek V4.1 model package naming (#56741)

2026-09-13 23:17:50 -07:00

triton_utils

[Bugfix][CPU] Fix several bugs (#54042)

2026-08-31 15:33:06 +08:00

usage

[Bugfix] Close usage telemetry HTTP sessions (#51219)

2026-08-08 00:29:36 +00:00

utils

[Fast Start] Support fp4 (#55465)

2026-09-10 13:22:17 +00:00

v1

[Bugfix][KV Offload] Reuse in-flight async lookup probes (#55823)

2026-09-14 15:02:45 +03:00

vllm_flash_attn

[Warmup] Gemma 4 de-JITification (#55768)

2026-09-13 16:47:21 -04:00

__init__.py

…

_aiter_ops.py

[ROCm][DSV4.1][Perf] Use AITER mHC for the delayed pre block (#56503)

2026-09-12 00:35:48 +00:00

_custom_ops.py

[3/N] HiSparse: host-resident sparse-MLA decode hot-buffering (#53781)

2026-09-12 00:36:47 -04:00

_xpu_ops.py

[XPU] Use fused_input_norm kernel in FusedInputNorm (#52945)

2026-09-07 13:44:31 +08:00

collect_env.py

[XPU] fix collecting oneccl version info (#47104)

2026-08-04 11:53:28 +00:00

connections.py

Reject oversized media before fully downloading it (#51896)

2026-08-24 01:15:09 +00:00

env_override.py

[ROCm][Bugfix] Fix profiler in TheRock image (#56190)

2026-09-10 13:39:33 -05:00

envs.py

[Frontend] Replace VLLM_ENABLE_SCALE_OUT_ENDPOINTS with --enable-scale-out (#55176)

2026-09-14 10:02:14 +00:00

exceptions.py

[Core] Add max_num_queued_reqs and max_num_queued_tokens for queue size management (#49445)

2026-08-31 14:26:49 +00:00

forward_context.py

[Feature] Support sequence parallel without the need for DP, 1.9%~5.0% E2E Throughput Improvement (#47070)

2026-07-05 05:41:30 -07:00

logger.py

[httpx migration] Import httpx from huggingface_hub (#56460)

2026-09-11 16:29:31 +01:00

logits_process.py

…

logprobs.py

[Bugfix] Handle empty FlatLogprobs slices and delta output (#53704)

2026-08-26 21:00:54 +08:00

model_inspection.py

…

outputs.py

[Frontend][Core][Spec Decode] Per-request acceptance stats in OpenAI API responses (#48915)

2026-08-20 15:42:49 -07:00

pooling_params.py

[MISC] Cleanup deprecated parameters (#53559)

2026-08-24 10:19:04 -07:00

py.typed

…

sampling_params.py

[Bugfix][Beam Search] Respect skip_special_tokens during decoding (#56211)

2026-09-14 05:49:44 +00:00

scalar_type.py

Humming support for 2/3/5/6/7-bit pack-quantized weight-only inference (#46389)

2026-06-24 13:53:54 -04:00

scripts.py

…

sequence.py

…

tasks.py

[Frontend] Move api_server.py out openai folder (#52131)

2026-08-19 09:43:46 +00:00

version.py

…