I want to reproduce Open R1 on qwen2.5 7b models， Could you share the specific traning YAML ?

[zhangyuanscall]

Could you please provide the training parameters used to reproduce Open R1? Could you share the specific YAML file?

[xiaozhang23957]

@zhangyuanscall Thank you for your interest in our work! The training YAML configuration and training command are as follows:

https://github.com/AMAP-ML/GPG/blob/main/open-r1/recipes/Qwen2.5-Math-7B/grpo/config_simple_rl_math_l35_v1.yaml

Please refer：https://github.com/AMAP-ML/GPG/blob/main/open-r1/train.sh