Ferrebo 912744d066

[Fix] optimize visual token mask with caching and multi-token support (#28374) ...

Signed-off-by: Ferrebo <itachi971009@gmail.com>
Signed-off-by: kebo01 <kebo01@baidu.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

2025-11-10 13:23:49 +00:00

..

assets

[VLM] Add Qwen3-VL generation test (#25185)

2025-10-29 12:19:37 +00:00

attention

[V0 deprecation] Remove no longer used get_metadata_cls (#28370)

2025-11-10 14:32:09 +08:00

benchmarks

[Feature] Default ignore_eos True for random dataset (#28227)

2025-11-07 07:35:33 -05:00

compilation

remove resolve_op_overloads and use splitting_ops directly (#28081)

2025-11-08 01:13:13 +00:00

config

[DCP] Support dcp kv_cache interleave size > 1 (#26696)

2025-11-09 04:45:27 +09:00

device_allocator

[Chore] Separate out vllm.utils.platform_utils.py (#27374)

2025-10-23 19:08:06 +00:00

distributed

[Core] Simplify async KV output aggregation (#28327)

2025-11-09 09:44:13 -08:00

engine

[DCP] Support dcp kv_cache interleave size > 1 (#26696)

2025-11-09 04:45:27 +09:00

entrypoints

[Feat] Drop-in Torch CUDA Profiler (#27841)

2025-11-08 14:07:37 -08:00

inputs

…

logging_utils

…

lora

[Bugfix][LoRA][Spec Decode] Support LoRA with speculative decoding (#21068)

2025-11-08 01:58:22 +00:00

model_executor

[Fix] optimize visual token mask with caching and multi-token support (#28374)

2025-11-10 13:23:49 +00:00

multimodal

[Core][MM] Add mechanism to configure multimodal fields which should stay on CPU (#28168)

2025-11-07 12:14:29 +00:00

platforms

[ROCm][Platform] Add RX7900XTX device id in _ROCM_DEVICE_ID_NAME_MAP (#28279)

2025-11-09 23:09:36 +00:00

plugins

[Frontend][4/N] Improve all pooling task | Add plugin pooling task (#26973)

2025-10-23 14:46:18 +00:00

profiler

[Feat] Drop-in Torch CUDA Profiler (#27841)

2025-11-08 14:07:37 -08:00

ray

…

reasoning

reasoning_content -> reasoning (#27752)

2025-11-08 12:15:08 +00:00

third_party

…

transformers_utils

reasoning_content -> reasoning (#27752)

2025-11-08 12:15:08 +00:00

triton_utils

…

usage

[Misc] Add TPU usage report when using tpu_inference. (#27423)

2025-10-23 20:29:37 -07:00

utils

[flashinfer][fix] do not check nvcc availability when using pre-downloaded cubins (#27990)

2025-11-07 22:25:21 -08:00

v1

[Metrics] Refactor LoRA state tracking (#26801)

2025-11-10 16:34:36 +08:00

vllm_flash_attn

…

__init__.py

…

_bc_linter.py

…

_custom_ops.py

[Perf][DeepSeek] Add sigmoid+bias fusion to fused_grouped_topk from TRTLLM (#28124)

2025-11-07 18:20:55 -08:00

_ipex_ops.py

[Multimodal][XPU]Enable vision attn backend for xpu platform (#27525)

2025-11-01 04:45:02 +00:00

beam_search.py

…

collect_env.py

…

connections.py

…

env_override.py

[Graph Partition][Cache] Use inductor partition ops config (#27702)

2025-11-05 13:04:48 +00:00

envs.py

[ROCm] Add env to enable/disable aiter triton gemm (#28321)

2025-11-08 22:27:00 -08:00

forward_context.py

…

logger.py

[Log] Optimize Startup Log (#26740)

2025-10-24 19:27:04 -04:00

logits_process.py

…

logprobs.py

[Perf] Introduce FlattenLogprobs to store logprobs results to reduce GC overhead (#28171)

2025-11-07 00:27:12 -08:00

outputs.py

[Model] Add num_cached_tokens for PoolingRequestOutput (#27378)

2025-10-23 14:03:42 +08:00

pooling_params.py

[Frontend][Doc][5/N] Improve all pooling task | Polish encode (pooling) api & Document. (#25524)

2025-10-30 12:13:05 +00:00

py.typed

…

sampling_params.py

…

scalar_type.py

…

scripts.py

…

sequence.py

…

tasks.py

[Frontend][4/N] Improve all pooling task | Add plugin pooling task (#26973)

2025-10-23 14:46:18 +00:00

tracing.py

…

version.py

…