kevin 5d42f19e0a [BugFix][Scheduler] Fix can_schedule_block_num_threshold calculation (#6541) ...

* fix mtp acceptance rate decline

* [BugFix][Scheduler] Fix can_schedule_block_num_threshold calculation

Fix the calculation of can_schedule_block_num_threshold in
ResourceManagerV1. The original formula using need_prefill_tokens
could lead to incorrect threshold values. Now directly use
num_chunk_new_block for accurate block scheduling.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

2026-02-28 16:23:18 +08:00

..

sched

[BugFix][Scheduler] Fix can_schedule_block_num_threshold calculation (#6541)

2026-02-28 16:23:18 +08:00

__init__.py

polish code with new pre-commit rule (#2923)

2025-07-19 23:19:27 +08:00

args_utils.py

[Optimization] Enable BF16 gate computation for GLM and Qwen (#6457)

2026-02-26 21:08:46 -08:00

async_llm.py

[Optimization] The pre- and post-processing pipeline do not perform dict conversion (#5494)

2026-01-22 00:50:52 +08:00

common_engine.py

[Optimization] Enable BF16 gate computation for GLM and Qwen (#6457)

2026-02-26 21:08:46 -08:00

engine.py

[Optimization] Enable BF16 gate computation for GLM and Qwen (#6457)

2026-02-26 21:08:46 -08:00

expert_service.py

[Others] Exit to ensure no residual processes (cpu cache & dp) (#6377)

2026-02-09 20:38:38 +08:00

kv_cache_interface.py

bug: fix list to List (#4818)

2025-11-06 16:13:12 +08:00

pooling_params.py

[Feature] support reward model (#5301)

2025-12-02 14:55:31 +08:00

request.py

[RL] add pause, update_weights, resume interface for async RL (#6052)

2026-01-23 10:18:07 +08:00

resource_manager.py

[Feature] Support stopping the inference for the corresponding request in the online service after a disconnection request. (#5320)

2026-01-16 11:46:13 +08:00

sampling_params.py

[Feature] Add Deterministic Inference Support (#6476)

2026-02-26 19:31:51 -08:00

tasks.py

[feature] support reward api (#4518)

2025-10-29 00:20:28 +08:00