包package 2.x

agentscope-extensions-training

AgentScope Java - Extensions - Training · 2.0.4

AgentScope Extensions - Training Data Collection and Export

groupIdGroupId	io.agentscope
版本Version	2.0.4
版本线Line	2.x
分类Category	extension
包Packages	5
类型Types	18 top-level / 26 total

包Packages

`io.agentscope.core.training.backend`	2
`io.agentscope.core.training.backend.dto`	3
`io.agentscope.core.training.reward`	1
`io.agentscope.core.training.runner`	6
`io.agentscope.core.training.strategy`	6

类型Types

类型Type	种类Kind	包Package	说明Description
`TrinityClient`	类class	`io.agentscope.core.training.backend`	Trinity 客户端，封装与远端服务的连接和调用。 Trinity Backend Client
`TrinityModelAdapter`	类class	`io.agentscope.core.training.backend`	Trinity 模型适配器，对智能体行为进行评测、追踪与可观测性分析。 Trinity Model Adapter
`CommitRequest`	类class	`io.agentscope.core.training.backend.dto`	提交请求，对智能体行为进行评测、追踪与可观测性分析。 Trinity Commit API Request
`FeedbackRequest`	类class	`io.agentscope.core.training.backend.dto`	Feedback 请求，对智能体行为进行评测、追踪与可观测性分析。 Trinity Feedback API Request
`StatusResponse`	类class	`io.agentscope.core.training.backend.dto`	状态响应，对智能体行为进行评测、追踪与可观测性分析。 Trinity Common Status Response
`RewardCalculator`	接口interface	`io.agentscope.core.training.reward`	定义 Reward Calculator契约的接口。 Reward Calculator Interface
`AgentCloner`	类class	`io.agentscope.core.training.runner`	智能体 Cloner，对智能体行为进行评测、追踪与可观测性分析。 Agent Cloning Utility Class
`RunExecutionContext`	类class	`io.agentscope.core.training.runner`	运行执行上下文，对智能体行为进行评测、追踪与可观测性分析。 Run Execution Context
`TaskExecutionRegistry`	类class	`io.agentscope.core.training.runner`	任务执行注册表，对智能体行为进行评测、追踪与可观测性分析。 Task Execution Registry
`TrainingConfig`	类class	`io.agentscope.core.training.runner`	Training 配置，保存相关配置项。 Configuration for Training Runner.
`TrainingRouter`	类class	`io.agentscope.core.training.runner`	Training 路由器，对智能体行为进行评测、追踪与可观测性分析。 Training Router Hook
`TrainingRunner`	类class	`io.agentscope.core.training.runner`	Training 运行器，对智能体行为进行评测、追踪与可观测性分析。 Training Runner - Training Process Controller
`ExplicitMarkingStrategy`	类class	`io.agentscope.core.training.strategy`	Explicit Marking Strategy，对智能体行为进行评测、追踪与可观测性分析。 Explicit Marking Strategy - Selection strategy based on user explicit marking
`SamplingRateStrategy`	类class	`io.agentscope.core.training.strategy`	Sampling 速率 Strategy，对智能体行为进行评测、追踪与可观测性分析。 Sampling Rate Strategy - Selection strategy based on sampling rate
`SelectionDecision`	类class	`io.agentscope.core.training.strategy`	Selection 决策，对智能体行为进行评测、追踪与可观测性分析。 Selection Decision - Selection decision result
`TrainingAnnotation`	类class	`io.agentscope.core.training.strategy`	Training 注解，对智能体行为进行评测、追踪与可观测性分析。 TrainingAnnotation represents a training annotation for an agent call.
`TrainingContext`	类class	`io.agentscope.core.training.strategy`	Training 上下文，对智能体行为进行评测、追踪与可观测性分析。 TrainingContext provides API for marking agent calls for training.
`TrainingSelectionStrategy`	接口interface	`io.agentscope.core.training.strategy`	定义 Training Selection Strategy契约的接口。 Training Selection Strategy - Training request filtering strategy