Back to Home
RadarAI··Papers & Tech

Microsoft 与 Hugging Face 用 ThinkingBox 把数据库评分变成 RL 奖励训练 Agent

中文摘要

Microsoft与Hugging Face发布ThinkingBox-Training,利用数据库检查作为RL奖励训练智能体,代码已开源。

English Summary

Microsoft and Hugging Face released ThinkingBox-Training, using database checks as RL rewards for agents. Code is open-source.

Original Excerpt

Microsoft 与 Hugging Face 发布 ThinkingBox-Training,将智能体任务终态数据库的可执行检查用作 RL 奖励,代码已在 GitHub 开源。