在线训练

<dependency>
    <groupId>io.agentscope</groupId>
    <artifactId>agentscope-extensions-training</artifactId>
    <version>${agentscope.version}</version>
</dependency>

import io.agentscope.core.training.runner.TrainingRunner;
import io.agentscope.core.training.strategy.SamplingRateStrategy;

TrainingRunner runner = TrainingRunner.builder()
    .trinityEndpoint("http://localhost:8080")
    .modelName("/path/to/model")
    .selectionStrategy(SamplingRateStrategy.of(0.1))   // 10% 流量进训练
    .rewardCalculator(agent -> 0.0)                    // 自定义奖励
    .commitIntervalSeconds(300)                        // 每 5 分钟 commit 一次
    .build();

runner.start();          // 拦截 Agent，开始采样

// 业务侧照常使用 Agent，无须感知 runner
agent.call(msg).block();

runner.stop();           // 停止训练流水线

TrainingRunner runner = TrainingRunner.builder()
    .trinityEndpoint(endpoint)
    .modelName(model)
    .selectionStrategy(SamplingRateStrategy.of(0.1))
    .rewardCalculator(new MyMetricRewardCalculator())
    .build();