wehub-resource-sync
50 lines
1.4 KiB
Python
50 lines
1.4 KiB
Python
"""
|
|
Task evaluation module.
|
|
|
|
This module provides the task system for bench_env:
|
|
- JudgeInput/JudgeResult: Input/output types for evaluation
|
|
- BaseTask/BaseApp: Abstract base classes
|
|
- TaskSampler: Parameter sampling from environment state
|
|
- CriteriaTask/AnswerTask: Common task patterns
|
|
- TaskRegistry: Registry for discovering task classes
|
|
|
|
App-specific modules:
|
|
- bench_env.task.wechat: WeChat tasks and state accessor
|
|
- bench_env.task.redbook: Redbook tasks and state accessor
|
|
"""
|
|
|
|
from bench_env.task.judge import JudgeInput, JudgeResult
|
|
from bench_env.task.base import BaseTask, BaseApp
|
|
from bench_env.task.sampler import TaskSampler, SampleResult
|
|
from bench_env.task.common_tasks import CriteriaTask, AnswerTask
|
|
from bench_env.task.registry import TaskRegistry, load_tasks
|
|
|
|
# App-specific exports (for convenience)
|
|
from bench_env.task.wechat import Wechat
|
|
from bench_env.task.file_manager import FileManager, FileSystem
|
|
from bench_env.task.redbook import Redbook
|
|
|
|
__all__ = [
|
|
# Judge types
|
|
"JudgeInput",
|
|
"JudgeResult",
|
|
# Task base classes
|
|
"BaseTask",
|
|
"BaseApp",
|
|
# Sampler
|
|
"TaskSampler",
|
|
"SampleResult",
|
|
# Common task patterns
|
|
"CriteriaTask",
|
|
"AnswerTask",
|
|
# Registry
|
|
"TaskRegistry",
|
|
# Convenience function
|
|
"load_tasks",
|
|
# App-specific state accessors
|
|
"FileManager",
|
|
"FileSystem",
|
|
"Wechat",
|
|
"Redbook",
|
|
]
|