refactor: TaskManager 剥离页管理,STOP 移至 task.py

- TaskManager 移除 page_cache/page_size 依赖,增 pull_candidates/activate/return_to_waiting
- Executor 增 allocate_pages_for_activation/free_task_pages,承接全部页操作
- STOP 从 cache.py 移至 task.py
- scheduler loop 显式装配: 清理→释页 / 拉取→分配→激活
- sampling.py → sample.py
This commit is contained in:
2026-05-11 14:04:31 +08:00
parent 317ed90bac
commit 73d6cc0f26
7 changed files with 78 additions and 79 deletions
+3 -2
View File
@@ -13,7 +13,7 @@ from astrai.inference.engine import (
GenerationRequest,
InferenceEngine,
)
from astrai.inference.sampling import (
from astrai.inference.sample import (
BaseSamplingStrategy,
SamplingPipeline,
TemperatureStrategy,
@@ -22,7 +22,7 @@ from astrai.inference.sampling import (
sample,
)
from astrai.inference.scheduler import InferenceScheduler
from astrai.inference.task import Task, TaskStatus
from astrai.inference.task import STOP, Task, TaskStatus
__all__ = [
# Engine / Requests
@@ -31,6 +31,7 @@ __all__ = [
"GenerationParams",
# Scheduler
"InferenceScheduler",
"STOP",
"Task",
"TaskStatus",
# Sampling (Strategy pattern)