Taiji: Pareto Optimal Policy Optimization with Semantics-IDs Trade-off for Industrial LLM-Enhanced Recommendation
Taiji: 面向工业LLM增强推荐的帕累托最优策略优化与语义ID权衡
机构 * Kuaishou Technology(快手科技) ; Unaffiliated(无隶属)
AI总结 提出Taiji框架,通过逆向工程推理和开放拒绝采样生成高质量CoT数据,并采用帕累托最优策略优化(POPO)自适应调整跨域奖励权重,实现LLM语义知识与推荐ID特征的帕累托最优权衡,在快手广告平台部署后服务超4亿日活用户。
Comments 8 pages, 2 figures