search

AliaksandrSiarohin / first-order-model

This repository contains the source code for the paper First Order Motion Model for Image Animation
https://aliaksandrsiarohin.github.io/first-order-model-website/
MIT License
14.57k stars 3.23k forks source link

Not Training on custom dataset #465

Open sauravns opened 3 years ago

sauravns commented 3 years ago

I was trying to train on a custom dataset, it is not showing any progress even after training for 12hrs. I have used the same dataset for monkey net, and it worked fine, so couldn't figure out what the issue is. It just shows a few warnings:

"/test/first-order-model/env/lib/python3.7/site-packages/torch/nn/functional.py:2423: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details. "See the documentation of nn.Upsample for details.".format(mode)) /test/first-order-model/env/lib/python3.7/site-packages/torch/nn/functional.py:1332: UserWarning: nn.functional.sigmoid is deprecated. Use torch.sigmoid instead. warnings.warn("nn.functional.sigmoid is deprecated. Use torch.sigmoid instead.") /test/first-order-model/env/lib/python3.7/site-packages/torch/nn/parallel/_functions.py:61: UserWarning: Was asked to gather along dimension 0, but all input tensors were scalars; will instead unsqueeze and return a vector. warnings.warn('Was asked to gather along dimension 0, but all '

and the training process is not progressing as the log file was empty. Can someone help me out.

AliaksandrSiarohin commented 3 years ago

Set num_repeats to 1.

sauravns commented 3 years ago

yea Its working now. Thanks. what does num_repeats represent?

AliaksandrSiarohin commented 3 years ago

Some dataset have short epochs only few steps, to make epochs longer this parameter is used.