Enhancing SFT Training Efficiency Using Packing and FlashAttention2 with Position IDs#31629
Merged
ArthurZucker merged 13 commits intohuggingface:mainfrom RhuiDih:dev/fa_packing_posidJul 23, 2024
+226
Commits
Commits on Jul 23, 2024
authored andRhui Dih LeecommittedRhui Dih Lee- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee
- authored andRhui Dih LeecommittedRhui Dih Lee