Reproduction and Reward-related Issues #1
Description
I want to express my appreciation for your impressive work on the Text2Touch project. The research and the techniques presented are truly inspiring, and I am excited to attempt reproducing the results for further exploration.
However, I have encountered a challenge during my reproduction of the work. Specifically, while using the large model (deepseek) for reward training, I am unable to find an effective reward to proceed with. I wanted to inquire if there are any reward-related files or configurations that align with the results presented in your paper, which I could use to directly facilitate the completion of my reproduction efforts.
I greatly appreciate your time and any guidance you could provide on this matter.Thank you once again for your outstanding contribution to the field!