Even with reinforcement learning as strong as OpenAI's, with their incredible … · @Chonghan_Liu · TwKit

@Chonghan_Liu

Even with reinforcement learning as strong as OpenAI's, with their incredible RL infra built by people like Jiayi Weng, and TRPO and PPO authors on the team, they still run into issues with getting hacked. It's an ongoing battle. And I think the hardest part in all of this is reviewing the entire trajectory.

曝光 625 · 评论 1 · 点赞 10 · 书签 2