[trainer] feat: add per-round logprob mismatch metrics for multi-turn training#5229
Open
aoshen524 wants to merge 1 commit intoverl-project:mainfrom
Open
[trainer] feat: add per-round logprob mismatch metrics for multi-turn training#5229aoshen524 wants to merge 1 commit intoverl-project:mainfrom
aoshen524 wants to merge 1 commit intoverl-project:mainfrom