Questions about the chit chat benchmark in RLVER

Excellent project! Your paper is really instructive to us!

I found that your fine-tuned models' performance in chit chat is not particularly impressive. But I think, in fact, such everyday scenarios might be more common for chatbots. I would like to know if you have tried training on more chit chat scenarios? Also, I would like to know if you are willing to open-source the chit chat test data?