Woosuk Kwon 5fcb0cdd68 [Model Runner V2] Use FP32 for Gumbel Noise (#34854) ...

Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>

2026-02-18 17:07:37 -08:00

..

gpu

[Model Runner V2] Use FP32 for Gumbel Noise (#34854)

2026-02-18 17:07:37 -08:00

__init__.py

[V1] Implement vLLM V1 [1/N] (#9289)

2024-10-22 01:24:07 -07:00

block_table.py

[V1][Hybrid] Mamba Prefix Caching with align mode (#30877)

2026-01-23 09:56:48 -08:00

cp_utils.py

[V1][Hybrid] Mamba Prefix Caching with align mode (#30877)

2026-01-23 09:56:48 -08:00

cpu_model_runner.py

[Feature] OTEL tracing during loading (#31162)

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

cpu_worker.py

[Core] Profiler improvements and lazy initialization (#33198)

2026-02-12 16:16:38 -08:00

dp_utils.py

[feature] extend DBO to XBO (#30120)

2025-12-16 00:04:01 -05:00

ec_connector_model_runner_mixin.py

[Misc] Introduce ec_both role EC (encoder cache) connector (#34182)

2026-02-10 18:55:35 +00:00

gpu_input_batch.py

[V1][Hybrid] Mamba Prefix Caching with align mode (#30877)

2026-01-23 09:56:48 -08:00

gpu_model_runner.py

[Feature][Perf] Support Selective CPU Weight Offloading (#34535)

2026-02-13 20:02:24 -08:00

gpu_ubatch_wrapper.py

[Bugfix] Fix ubatch wrapper num_tokens calculate (#33694)

2026-02-04 16:41:45 +00:00

gpu_worker.py

[Feature] Pipeline Parallel Async send/recv, 2.9% E2E throughput improvement (#33368)

2026-02-13 16:38:16 +08:00

kv_connector_model_runner_mixin.py

[Cleanup] Remove unused KVConnectorModelRunnerMixin methods (#32077)

2026-01-21 11:16:37 +08:00

lora_model_runner_mixin.py

Reduce the kernel overhead when num of active loras is smaller than max loras. Multiple cuda graphs are captured for each num of active-loras. (#32005)

2026-02-02 12:30:06 -05:00

mamba_utils.py

[Hybrid] Enable mamba prefix cache "align" mode with async scheduling (#33997)

2026-02-14 13:15:56 -08:00

tpu_input_batch.py

[Cleanup] Remove unused ModelRunner V1 InputBatch.num_tokens field (#30218)

2025-12-18 09:17:00 -08:00

ubatch_utils.py

[6/N][Attention] Move utils to more appropriate locations (#32215)

2026-01-13 05:38:52 -08:00

ubatching.py

[Misc] Replace Optional[X] with X | None syntax (#33332)

2026-01-30 01:56:59 -08:00

utils.py

[Refactor] Move profiling methods to MM budget (#33559)

2026-02-02 23:27:00 +08:00

worker_base.py

[Feature] OTEL tracing during loading (#31162)

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

workspace.py

[Misc] Replace Optional[X] with X | None syntax (#33332)

2026-01-30 01:56:59 -08:00

xpu_model_runner.py

Replace torch.cuda.Event with torch.Event for better hardware compatibility (#26985)

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

xpu_worker.py

[XPU][1/N] Deprecate ipex and switch to vllm-xpu-kernels for xpu platform (#33379)

2026-02-02 22:46:10 -08:00