vllm/vllm at d886c26d4d4fef7d079696beb4ece1cfb4b008a8 - vllm

assets

[Frontend] Remove librosa from audio dependency (#37058 )

2026-03-21 11:36:15 +08:00

benchmarks

[UX] Defer some imports on CLI paths to save ~2s (#40056 )

2026-04-16 19:48:37 -07:00

compilation

[compile] Invoke split FX graph by codegen. (#38657 )

2026-04-15 21:03:41 -07:00

config

Add nvfp4 support to reshape_and_cache_flash (#37332 )

2026-04-17 07:28:00 -07:00

device_allocator

[CI] Fix mypy for vllm/device allocator (#35518 )

2026-03-02 15:53:18 +00:00

distributed

[KV Offload] Pass request context (#39185 )

2026-04-19 08:54:59 +03:00

engine

[Attention Backend] TurboQuant: 2-bit KV cache compression with 4x capacity (#38479 )

2026-04-14 19:57:13 -07:00

entrypoints

[Frontend] Preserve structured output special tokens in offline LLM.chat (#39352 )

2026-04-18 19:46:07 -04:00

inputs

[Misc] Reorganize inputs (#35182 )

2026-03-25 10:22:54 -07:00

ir

[vLLM IR] rework gemma_rms_norm (#39014 )

2026-04-07 01:37:00 -07:00

kernels

[Kernel] [Helion] Force disable HOP path due to performance regression (#40171 )

2026-04-17 17:36:49 -04:00

logging_utils

[vLLM IR] 1/N Implement IR skeleton and rms_norm op (#33825 )

2026-03-31 22:15:05 -04:00

lora

[BugFix][XPU] fix lora ops bgmv_expand size not match (#39989 )

2026-04-20 08:24:50 +08:00

model_executor

Fix MoE backend selection for LoRA (unquantized MoE) (#40273 )

2026-04-19 17:18:40 +00:00

multimodal

[Refactor] Drop direct dependency on librosa (#39079 )

2026-04-18 06:55:38 +00:00

parser

[Bugfix] Accept **kwargs in MiniMaxM2Parser.__init__() (#39861 )

2026-04-16 15:18:32 +08:00

platforms

[ZenCPU] AMD Zen CPU Backend with supported dtypes via zentorch weekly (#39967 )

2026-04-18 06:22:37 +00:00

plugins

[Frontend][4/n] Improve pooling entrypoints | pooling. (#39153 )

2026-04-09 10:09:45 +00:00

profiler

[BugFix][CPU] Add CPU profiler summary file output (#38366 )

2026-04-10 13:41:15 +08:00

ray

[Ray] Propagate third-party env vars to Ray workers via prefix matching (#34383 )

2026-02-17 01:08:42 -08:00

reasoning

[Misc][UX] Map mimo reasoning and tooling parsers (#40089 )

2026-04-17 16:49:21 +00:00

renderers

[Bugfix] Use dedicated MM processor cache in /tokenize to prevent sender-cache pollution (#38545 )

2026-04-01 21:14:49 -07:00

third_party

[FlashMLA] Update FlashMLA to expose new arguments (#32810 )

2026-01-21 22:02:39 -07:00

tokenizers

[Mistral Grammar] Fix tool and reasoning parsing (#39217 )

2026-04-15 21:05:04 -07:00

tool_parsers

[Bugfix] Kimi-K2 tool parser streaming - fix token leakage, argument truncation, and content dropping (#38579 )

2026-04-19 01:30:27 -07:00

tracing

[Feature] OTEL tracing during loading (#31162 )

2026-02-05 16:59:28 -08:00

transformers_utils

Optimize nemotron VL image/video preprocessing (#40283 )

2026-04-19 15:06:20 +00:00

triton_utils

Add XPU MLA Sparse backend for DeepSeek v3.2 (#33230 )

2026-03-11 19:19:15 +08:00

usage

replace cuda_device_count_stateless() to current_platform.device_count() (#37841 )

2026-03-31 22:32:54 +08:00

utils

[Frontend] Add multimodal support to /inference/v1/generate endpoint (#38405 )

2026-04-17 20:31:56 -07:00

v1

[ROCm] Fix cu_seqlens_q off-by-one in AITER FA speculative decode path (#39120 )

2026-04-19 18:34:33 +00:00

vllm_flash_attn

[Attention] relax the head dim 512 and paged kv for sm90+FA4 (#38835 )

2026-04-08 18:23:18 +00:00

__init__.py

Remove bc-lint (#35274 )

2026-02-26 03:01:01 -08:00

_aiter_ops.py

Fix #33773 : Replace unconditional pandas import with PlaceholderModule (#39990 )

2026-04-16 02:06:51 -07:00

_custom_ops.py

[Kernel] Add MXFP4 W4A4 CUTLASS MoE kernel for SM100 (#37463 )

2026-04-17 16:42:32 -07:00

_oink_ops.py

[Perf] Add opt-in SM100 Oink RMSNorm custom-op path (#31828 )

2026-02-24 23:01:53 -08:00

_xpu_ops.py

[XPU]fake impl for xpu fp8_gemm (#39984 )

2026-04-18 08:53:56 +08:00

beam_search.py

[Misc] Reorganize inputs (#35182 )

2026-03-25 10:22:54 -07:00

collect_env.py

[XPU]Enhance environment collection for Intel XPU and optimize layout (#35698 )

2026-04-13 12:51:46 +08:00

connections.py

[CI] Add retry with 4x backoff to HTTP fetches for transient failures (#37218 )

2026-03-19 19:00:20 +00:00

env_override.py

[UX] Defer some imports on CLI paths to save ~2s (#40056 )

2026-04-16 19:48:37 -07:00

envs.py

[MoE Refactor] Remove MoE DP chunking (#39107 )

2026-04-14 09:48:05 -04:00

exceptions.py

[openapi server] log exception in exception handler(2/N) (#36201 )

2026-03-10 20:16:30 -07:00

forward_context.py

[MoE Refactor] Remove MoE DP chunking (#39107 )

2026-04-14 09:48:05 -04:00

logger.py

[Log] Log once in local node by default (#37568 )

2026-03-19 12:04:59 -07:00

logits_process.py

[Misc] Refactor tokenizer interface (#29693 )

2025-11-29 04:02:21 -08:00

logprobs.py

Make mypy opt-out instead of opt-in (#33205 )

2026-01-29 09:12:26 +00:00

model_inspection.py

[UX] Add vLLM model inspection view (#29450 )

2026-01-09 10:12:35 -07:00

outputs.py

[V0 Deprecation] Remove unused MM placeholders in request output (#34944 )

2026-02-20 06:19:23 -08:00

pooling_params.py

[Model] Deprecate the score task (this will not affect users). (#37537 )

2026-03-20 08:07:56 +00:00

py.typed

…

sampling_params.py

[Mistral Grammar] Fix tool and reasoning parsing (#39217 )

2026-04-15 21:05:04 -07:00

scalar_type.py

…

scripts.py

…

sequence.py

[Model Runner V2][Minor] Simplify PP logic (#38031 )

2026-03-24 13:57:17 -07:00

tasks.py

[Model] Deprecate the score task (this will not affect users). (#37537 )

2026-03-20 08:07:56 +00:00

version.py

…