Yukang

/

Qwen2.5-7B-Open-R1-GRPO

like 0

I am wondering which version of open-r1/OpenR1-Math-220k do we use during this GPRO process? I review the dataset open-r1/OpenR1-Math-220k and I found it has a "default" version of data and a "extended" version of data. I am not sure which version do we use here.