How do I finetune this model and what is the performance difference from this implementation versus the official one?
#6
by
SS12444
- opened
Hi, thanks for the work. Is there a comparison on the benchmark scores after adding this implementation vs the original one? I also have another question which is can we use PEFT method to finetune this model (is there an example script)? Or how can we finetune it without converting the weights?
Thank you.