org.noear.solon.ai.agent.react

类 ReActOptionsAmend

java.lang.Object

org.noear.solon.ai.chat.ModelOptionsAmend<ReActOptionsAmend,ReActInterceptor>

org.noear.solon.ai.agent.react.ReActOptionsAmend

public class ReActOptionsAmend
extends ModelOptionsAmend<ReActOptionsAmend,ReActInterceptor>

ReAct 运行选项修正器（提供链式调用支持）

从以下版本开始:

3.8.1

字段概要

从类继承的字段 org.noear.solon.ai.chat.ModelOptionsAmend

autoToolCall, interceptors, skills, toolContext, tools

构造器概要

构造器

构造器和说明
ReActOptionsAmend(ReActOptions options)

方法概要

限定符和类型	方法和说明
ReActOptionsAmend	feedbackDescription(java.util.function.Function<ReActTrace,java.lang.String> provider)
ReActOptionsAmend	feedbackDescription(java.lang.String description)
ReActOptionsAmend	feedbackMode(boolean feedbackMode)
ReActOptionsAmend	feedbackReasonDescription(java.util.function.Function<ReActTrace,java.lang.String> provider)
ReActOptionsAmend	feedbackReasonDescription(java.lang.String description)
ReActOptionsAmend	maxSteps(int val) 设置单次任务最大推理步数
ReActOptionsAmend	maxStepsExtensible(boolean val)
ReActOptionsAmend	maxStepsLimit(int val)
ReActOptionsAmend	outputSchema(java.lang.String val)
ReActOptionsAmend	planningInstruction(java.util.function.Function<ReActTrace,java.lang.String> provider)
ReActOptionsAmend	planningInstruction(java.lang.String instruction)
ReActOptionsAmend	planningMode(boolean planningMode)
ReActOptionsAmend	retryConfig(int maxRetries, long retryDelayMs) 配置容错策略
ReActOptionsAmend	sessionWindowSize(int sessionWindowSize) 设置会话回溯窗口大小（控制短期记忆深度）

从类继承的方法 org.noear.solon.ai.chat.ModelOptionsAmend

autoToolCall, frequency_penalty, interceptorAdd, interceptorAdd, interceptorAdd, interceptors, isAutoToolCall, max_completion_tokens, max_tokens, option, optionRemove, options, optionSet, optionSet, presence_penalty, putAll, response_format, skillAdd, skillAdd, skillAdd, skills, temperature, tool_choice, tool, toolAdd, toolAdd, toolAdd, toolAdd, toolAdd, toolContext, toolContextPut, toolContextPut, tools, top_k, top_p, user, user

从类继承的方法 java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

构造器详细资料

ReActOptionsAmend

public ReActOptionsAmend(ReActOptions options)

方法详细资料

retryConfig

public ReActOptionsAmend retryConfig(int maxRetries,
                                     long retryDelayMs)

配置容错策略

参数:

maxRetries - 最大重试次数

retryDelayMs - 重试延迟（毫秒）

sessionWindowSize

public ReActOptionsAmend sessionWindowSize(int sessionWindowSize)

设置会话回溯窗口大小（控制短期记忆深度）

maxSteps

public ReActOptionsAmend maxSteps(int val)

设置单次任务最大推理步数

maxStepsLimit

public ReActOptionsAmend maxStepsLimit(int val)

maxStepsExtensible

public ReActOptionsAmend maxStepsExtensible(boolean val)

outputSchema

public ReActOptionsAmend outputSchema(java.lang.String val)

planningMode

public ReActOptionsAmend planningMode(boolean planningMode)

planningInstruction

public ReActOptionsAmend planningInstruction(java.lang.String instruction)

planningInstruction

public ReActOptionsAmend planningInstruction(java.util.function.Function<ReActTrace,java.lang.String> provider)

feedbackMode

public ReActOptionsAmend feedbackMode(boolean feedbackMode)

feedbackDescription

public ReActOptionsAmend feedbackDescription(java.lang.String description)

feedbackDescription

public ReActOptionsAmend feedbackDescription(java.util.function.Function<ReActTrace,java.lang.String> provider)

feedbackReasonDescription

public ReActOptionsAmend feedbackReasonDescription(java.lang.String description)

feedbackReasonDescription

public ReActOptionsAmend feedbackReasonDescription(java.util.function.Function<ReActTrace,java.lang.String> provider)