whu125

I downloaded your code. Did you not use subtitle information when testing Longvideobench?

Which repository should it be replaced with?

论文中提到Consistency training复用了SFT的数据，请问是完全一致的数据量吗？还是挑选了部分子集呢？ 同理，Router training具体使用了多少呢？

This is conducting RL training on the 0.4B model. Is my understanding correct?

1

The GPU usage is approximately 40GB, but during reproduction, I noticed that the loss is always 0, the inter-group advantages calculated are all 0, and the reward for each group...