[core / DDP] Fix RM trainer + DDP + quantization + propagate gradient_checkpointing_kwargs in SFT & DPO
#2024
| Job | Run time |
|---|---|
| 21s | |
| 6m 59s | |
| 9m 55s | |
| 6m 9s | |
| 10m 43s | |
| 8m 15s | |
| 9m 10s | |
| 51m 32s |