error when train my own data

[Linsongrong]

@XiaohangZhan Hello, I prepared my own data set according to the prompt, and then re-created config.yaml, there was an ‘out of memory’ error during training, and reducing the batchsize to 1 was still useless.

[XiaohangZhan]

Try to use pytorch 0.3.1. Note that this repo will not be maintained anymore.