Lianmin Zheng dc0705a504 Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

..

eplb_algorithms

Support picking variants of EPLB algorithms (#6728)

2025-05-29 08:12:01 -07:00

multimodal_processors

[Refactor] image data process in bench_serving (#6879)

2025-06-06 21:11:17 -07:00

cache_controller.py

fix bug that gpu0 occupies more memory when hicache is turned on (#5778)

2025-05-09 15:36:08 -07:00

configure_logging.py

Support penalty in overlap mode; return logprob with chunked prefill; improve benchmark scripts (#3988)

2025-03-03 00:12:04 -08:00

data_parallel_controller.py

feat: add direct routing strategy to DP worker (#6884)

2025-06-09 11:44:05 -07:00

detokenizer_manager.py

Support incremental streaming of logprob/token_ids between scheduler and detokenizer (#6225)

2025-05-12 14:33:38 -07:00

eplb_manager.py

Support layerwise rebalancing experts (#6851)

2025-06-05 00:05:52 -07:00

expert_distribution.py

Fix expert distribution dumping causes OOM (#6967)

2025-06-08 17:41:14 -07:00

expert_location_dispatch.py

Tiny update error hints (#6846)

2025-06-05 00:05:28 -07:00

expert_location.py

Support layerwise rebalancing experts (#6851)

2025-06-05 00:05:52 -07:00

io_struct.py

feat: add direct routing strategy to DP worker (#6884)

2025-06-09 11:44:05 -07:00

mm_utils.py

Refactor vlm embedding routine to use precomputed feature (#6543)

2025-05-24 18:39:21 -07:00

multimodal_processor.py

Clean up imports (#5467)

2025-04-16 15:26:49 -07:00

schedule_batch.py

Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

schedule_policy.py

[Minor] cleanup unused imports (#6358)

2025-05-16 14:52:38 -07:00

scheduler_output_processor_mixin.py

Support incremental streaming of logprob/token_ids between scheduler and detokenizer (#6225)

2025-05-12 14:33:38 -07:00

scheduler.py

Simplify prepare_extend_after_decode (#6987)

2025-06-09 16:39:21 -07:00

session_controller.py

Support precomputed multimodal features for Qwen-VL and Gemma3 models. (#6136)

2025-05-16 12:26:15 -07:00

tokenizer_manager.py

feat: add direct routing strategy to DP worker (#6884)

2025-06-09 11:44:05 -07:00

tp_worker_overlap_thread.py

Support incremental streaming of logprob/token_ids between scheduler and detokenizer (#6225)

2025-05-12 14:33:38 -07:00

tp_worker.py

Log if cuda graph is used & extend cuda graph capture to cuda-graph-max-bs (#6201)

2025-05-12 00:17:33 -07:00

utils.py

Correctly abort the failed grammar requests & Improve the handling of abort (#6803)

2025-06-01 19:00:07 -07:00