Ke Bao c28c536c91 Fix swa available memory check (#15867)

2025-12-26 13:19:01 +08:00

..

async_dynamic_batch_tokenizer.py

…

async_mm_data_processor.py

[VLM] Optimize async mm data process mechanism (#12066)

2025-11-01 01:24:53 +08:00

cache_controller.py

[BugFix] fix prefixcache performance and accuracy on ascend (#13573)

2025-12-08 02:16:20 +08:00

configure_logging.py

…

data_parallel_controller.py

DP: support piggyback server load report (#11469)

2025-12-25 11:35:05 +08:00

detokenizer_manager.py

Clean up the __init__ of TokenizerManager and DetokenizerManager (#15796)

2025-12-25 10:30:27 -08:00

disagg_service.py

…

io_struct.py

DP: support piggyback server load report (#11469)

2025-12-25 11:35:05 +08:00

mm_utils.py

Custom All Reduce for Piecewise Cuda Graph (#15356)

2025-12-25 23:55:15 +08:00

multi_tokenizer_mixin.py

[Feature] Enable return routed experts (#12162)

2025-12-21 15:16:43 +08:00

multimodal_processor.py

feat: support external custom models (#13429)

2025-11-20 11:16:58 +08:00

overlap_utils.py

[Feature] Xiaomi MiMo-V2-Flash day0 support (#15207)

2025-12-19 11:40:07 +08:00

request_metrics_exporter.py

Add RequestMetricsExporter utility to export request-level metrics (#10973)

2025-11-12 16:33:37 -08:00

schedule_batch.py

Add overlap scheduling for embeddings code path (#14032)

2025-12-24 18:24:18 -08:00

schedule_policy.py

bugfix[schedule]: Refactor sort method and add related UT (#13576)

2025-12-23 03:13:57 +08:00

scheduler_dp_attn_mixin.py

Add metrics for having prefill and decode in different ranks (#15752)

2025-12-24 21:35:35 +08:00

scheduler_enhancer.py

[scheduler] enhance scheduler in dp_attention mixed case with spec (#14201)

2025-12-14 02:52:26 +08:00

scheduler_input_blocker.py

…

scheduler_metrics_mixin.py

Fix prefill num tokens metrics (#15858)

2025-12-26 10:26:42 +08:00

scheduler_output_processor_mixin.py

DP: support piggyback server load report (#11469)

2025-12-25 11:35:05 +08:00

scheduler_pp_mixin.py

Improve engine customization interface (#15635)

2025-12-22 14:24:16 -08:00

scheduler_profiler_mixin.py

Tiny call cudaProfilerStart only on first rank in node (#14211)

2025-12-01 18:18:45 +08:00

scheduler_recv_skipper.py

[Feat]Add scheduler recv skipper weights to environment configuration (#13855)

2025-11-27 18:16:11 +08:00

scheduler_runtime_checker_mixin.py

Tiny refactor request logger (#15740)

2025-12-24 19:51:26 +08:00

scheduler_update_weights_mixin.py

Support sanity checking weight consistency especially for RL (#13854)

2025-11-27 20:25:12 +08:00

scheduler.py

Fix swa available memory check (#15867)

2025-12-26 13:19:01 +08:00

session_controller.py

[Test] Fix session control test (#12336)

2025-10-29 18:28:04 +08:00

template_manager.py

…

tokenizer_communicator_mixin.py

Tiny refactor request logger (#15740)

2025-12-24 19:51:26 +08:00

tokenizer_manager_multiitem_mixin.py

move multi-item scoring functions in tokenizer manager into a separate file (#14740)

2025-12-09 14:47:06 -08:00

tokenizer_manager.py

Clean up the __init__ of TokenizerManager and DetokenizerManager (#15796)

2025-12-25 10:30:27 -08:00

tp_worker.py

Adjust wrong mtp meaning introduce by mimo (#15632)

2025-12-23 02:06:46 +08:00

utils.py

Support EPLB balancedness prometheus metric without GPU->CPU synchronize (#15401)

2025-12-18 22:24:23 +08:00