fix: resolve OOM in long-sequence training via conditional entropy gradient tracking#1524
Open
ppraneth wants to merge 1 commit intoTHUDM:mainfrom
Open
fix: resolve OOM in long-sequence training via conditional entropy gradient tracking#1524ppraneth wants to merge 1 commit intoTHUDM:mainfrom
ppraneth wants to merge 1 commit intoTHUDM:mainfrom