Jay
Jay
me too
dear authors, how to solve it, thanks in advance
> Unfortunately, I don't know. I don't use `torch.amp` myself, and I last touched RevLib a while ago. Please let me know if you have a minimal script to reproduce...
> Do you have a minimal example to reproduce the error? Due to an important deadline recently, I'll try to give you a reply as soon as possible. Thanks for...
when loss.backward() warpped in torch.cuda.amp.autocast() context, this error is not reported.
Really thanks for your RevLib. I hope to contribute mysellf once I have enough time.
> Thanks for the feedback, I will check and update it in the next week. Thank you very much!
> Hi, I just update the spe part in > > https://github.com/fnzhan/RABIT/blob/658c6af2cbea1d6dbff87e769f7875250a47840b/data/pix2pix_dataset.py#L72 thank you!
how to get coarse level corr matrix R whose shape is [b,u,v,k] > https://github.com/fnzhan/RABIT/blob/658c6af2cbea1d6dbff87e769f7875250a47840b/models/networks/ranking_attention.py#L154