Which model do you load for downstream tasks? model_q or model_k?
codename1995 opened this issue · comments
Hi,
when I finetune your TARL pre-trained model on downsteam tasks, I notice that there are two models saved, model_q (online updated model) and model_k (momentum updated model). Which one do you load for down-stream tasks?
Hi! We load model_q you can check how we save it in this function:
Line 52 in 7f883d6
Thank you for your reply.
The code you quote is only used once at the begining of training. After that, the ckpts are saved by
Line 47 in 7f883d6
I think that is the reason why there are both model_q and model_k in the ckpts.