Closed wangkaihong closed 2 years ago
Hi, the model pretrained using synthetic data is under 'stage1', and the model under 'stage2' is our released model for testing. We conducted one iteration of pseudo-label training in our paper.
I see, thanks for the clarification!
Hi,
I notice that the pretrained model is under a folder named "stage2", does that mean the model is trained with 2 iterations of pseudo-label generation and self-training?
Thanks