Cognition· @cognition · X·· 3 小时前AI 评分34
AI 导读
我们的训练采用大规模 RL 进行。虽然我们对 SWE-2 的 RL 配方做了一些调整，但大部分收益来自更高质量、更多样化的 RL 任务。 我们的研究团队仔细审核了任务的指令、环境配置和验证器，同时还策划了若干新的任务类型。
正文
Our training proceeded with large-scale RL. While we made some tweaks to our RL recipe from SWE-2, most of the gains came from higher-quality, more diverse RL tasks.
Our research team carefully audited the instructions, environment setup, and verifiers for our tasks, while also curating several new task types.
来源：Cognition · x.com