logan7000/cogrpo-homo-qwen25-3b-math345-groupB
021
archive: wandb curve + provenance (run 0t02and7)
Add cogrpo-homo-qwen25-3b-math345-groupB best ckpt (eval_reward 0.664@s130)
initial commit
archive: wandb curve + provenance (run 0t02and7)
Add cogrpo-homo-qwen25-3b-math345-groupB best ckpt (eval_reward 0.664@s130)
initial commit