Jayant Parashar

Results 2 issues of Jayant Parashar

Which script is creating the processed data in pkl format? as used in data_utils.py file: CODEX_PREDICTIONS_PATH = 'processed_data/codex_predictions.pkl' I am trying to use your code as a baseline, please help.

In both the Trainers, Basic, and FSDP, there is an underlying pattern of GPU memory not being freed. Allocation keeps increasing in steps while utilization remains roughly constant. ![image](https://github.com/eric-mitchell/direct-preference-optimization/assets/31920414/53663685-e3dc-4118-a1df-f825c7b9374c) Does...