Catastrophic forgetting problem

[shatealaboxiaowang]

Hi dear:

Thanks for your open source, but when i finetuned (whatever full parameters or LoRa ) on my dataset, catastrophic forgetting kept coming up (decrease in performance on the humaneval), i do not know how to solve it, do you have any tops?

[guoday]

If you fine-tune on the Instruct model, I think such a phenomenon is normal. There are two reasons for this: one is that your data is inferior compared to the original data used to fine-tune the Instruct model, and the other reason is that further fine-tuning on the Instruct model could lead to overfitting.

[shatealaboxiaowang]

Thx, i will continue to check my data quality.