Skip to main content

New answers tagged

0 votes

RL Policy Gradient: How to deal with rewards that are strictly positive?

Neil's answer is good, except note that the gradient $\nabla_{\theta} log \pi_{\theta}(s_t,a_t)v_t$ is not always positive because $\nabla_{\theta} log \pi_{\theta}(s_t,a_t)$ is not always positive. ...
Lawrence Tang's user avatar
1 vote
Accepted

Reinforcement learning for ensemble models

I believe going through this link may help alleviate some of your concerns. https://superannotate.com/blog/rlhf-for-llm My two cents on this would be that you can essentially directly fine-tune the ...
xabash's user avatar
  • 86
0 votes

Interpretation of PPO learning curve, value loss, policy loss

Hard to say much without knowing the specifics of the problem setting, reward function. Seems like trying out a couple of different hyper-parameters may help you get better performance, for example ...
xabash's user avatar
  • 86

Top 50 recent answers are included