Fine-tuning a fine-tuned model

Hello, I’ve been attempting to fine-tune the previous model using new data. Fine-tuning has been completed successfully on my new dataset, and it is producing good results; however, the old model’s responses are affected, and it is producing poor results for older datasets.

I would appreciate any guidance on this.



Same here, been asking about this as well. It does not seem like it’s adding the new data-sets (prompt-completions) to the previous fine-tuned data sets. I’m not sure what’s occurring.