get_avg_embedding and idx in `PPO_discrete_main.py` line 164

[qin2mu4]

https://github.com/wang1351/Re2LLM/blob/fa46f696f48d72f3865d6f893fd76133e31cb69b/PPO_discrete_main.py#L164C8-L164C58

Hi, thank you for sharing your code. I was going through the implementation and noticed that in PPO_discrete_main.py, line 164:

self.state = get_avg_embedding(self.data[idx][0])

The function get_avg_embedding and the variable idx are not defined anywhere in the codebase. This makes it difficult to understand or reproduce the intended behavior of this part of the code.

Where should get_avg_embedding be defined? Was it omitted from the current version of the repository?

What is the intended value or role of idx in this context?

Providing this clarification would be very helpful for users trying to follow or reproduce your results. Thanks!

[wang1351]

Thank you for pointing it out. It should be something like torch.mean(torch.stack(embeddings_list)).numpy()