Woosuk Kwon 3e1ad40655

[Model Runner V2] Add apply_temperature option to gumbel_sample (#29276) ...

Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>

2025-11-23 14:13:00 -08:00

..

assets

…

attention

[Attention] Refactor FA block_size limitations to hybrid models only (#29084)

2025-11-22 06:38:44 -08:00

benchmarks

[Bugfix] Fix host and port join for ipv6 in bench serve (#28679)

2025-11-16 10:20:57 +00:00

compilation

[Bugfix]Fix a conditional to not check zero value (#28754)

2025-11-21 19:59:07 -08:00

config

[Misc] Further clean up chunked prefill and prefix caching init (#29186)

2025-11-22 19:34:15 +08:00

device_allocator

[Core][AMD] Migrate fully transparent sleep mode to ROCm platform (#12695)

2025-11-12 15:24:12 -08:00

distributed

[chore][LMCache connector] Remove useless logs from lmcache connector (#29069)

2025-11-22 03:18:00 +00:00

engine

[perf][cpu] Accelerate paged attention GEMMs (QK, PV) on Arm CPUs with NEON (#29193)

2025-11-22 09:04:36 -08:00

entrypoints

[Frontend][Responses API] Multi-turn (with type: "output_text") support for non-harmony requests (#29175)

2025-11-22 09:58:22 +00:00

inputs

[Bugfix][Perf] Revert applying HF processor on text-only inputs for multimodal models (#28858)

2025-11-17 14:49:25 +00:00

logging_utils

[Misc] Colorize logs (#29017)

2025-11-19 19:26:04 -05:00

lora

[LoRA] Cleanup FusedMoEWithLoRA (#29187)

2025-11-22 04:01:30 +00:00

model_executor

[Fix] Add SM check to flashinfer MOE backend (#29144)

2025-11-23 00:39:30 +00:00

multimodal

[Minor][Clean] Remove the legacy assertion in video (#29150)

2025-11-20 23:52:34 -08:00

platforms

[NIXL] Fix after virtual block_size for host_buffer with heter kv_layout (#29122)

2025-11-21 08:55:27 -08:00

plugins

[Doc] Update plugin doc (#28532)

2025-11-21 14:57:26 +00:00

profiler

[Log] Optimize startup log (#28948)

2025-11-21 08:46:20 -08:00

ray

[Small] Capture AttributeError when checking ray dependency. (#29024)

2025-11-20 22:54:10 -08:00

reasoning

…

third_party

…

transformers_utils

[Bugfix] Use lazy string reference for DeepseekV3Config in config registry (#28958)

2025-11-23 11:43:21 +00:00

triton_utils

…

usage

VLLM_USE_TRITON_FLASH_ATTN V0 variable deprecation (#27611)

2025-11-11 18:34:36 -08:00

utils

Revert "[Redo] #26368 (#28771)" (#29121)

2025-11-20 21:27:45 -08:00

v1

[Model Runner V2] Add apply_temperature option to gumbel_sample (#29276)

2025-11-23 14:13:00 -08:00

vllm_flash_attn

…

__init__.py

…

_aiter_ops.py

Upstream triton fp4 weight preshuffle (#28888)

2025-11-21 11:34:46 -05:00

_bc_linter.py

…

_custom_ops.py

[kernel][perf] support uncontiguous input for rms_norm kernel (#28103)

2025-11-20 19:39:09 -08:00

_ipex_ops.py

…

beam_search.py

…

collect_env.py

…

connections.py

…

env_override.py

[CI] Reorganize compile tests so new tests are automatically included in CI (#28625)

2025-11-19 06:13:50 -08:00

envs.py

[BugFix] Make sure to allocate worst case MoE workspace during profile run in the DP + EP case (#27426)

2025-11-21 11:41:52 -08:00

forward_context.py

[Performance][DeepGEMM] Estimate expected_m (#28694)

2025-11-15 13:52:14 +08:00

logger.py

[Misc] Colorize logs (#29017)

2025-11-19 19:26:04 -05:00

logits_process.py

…

logprobs.py

[Core] Switch Flat logprob control from environment variable to SamplingParams (#28914)

2025-11-19 02:10:02 +00:00

outputs.py

…

pooling_params.py

[Model] Allow users to control skip reading cache per request. (#28194)

2025-11-16 00:04:50 -08:00

py.typed

…

sampling_params.py

[V0 Deprecation] Remove best_of (#29090)

2025-11-21 11:40:40 +08:00

scalar_type.py

…

scripts.py

…

sequence.py

Fix IntermediateTensors initialization and add type hints (#28743)

2025-11-15 04:31:36 +00:00

tasks.py

…

tracing.py

…

version.py

…