Fine-tune pre-trained model with pruning

Hi,

I’m trying to fine-tune the pre-trained model with pruning and I have few queries on this. i.e., typically, How many epochs and data are needed to get performance same as conventional training?

and please share the links related to performance vs no of fine-tuning epochs vs database if any.

Thanks

HI there is no universial fine tuning schedule.

One popular setting is running the same epoch with the pre-trained model training. See https://arxiv.org/pdf/2104.08378.pdf for some reference.