Reproduced experiment results

[JalinWang]

Is there anyone who has reproduced the reported results? Does the code need to be modified?

Only 48% acc on: Leave-one-domain-out(photo), PACS official split, default parameters of ResNet50. Test result:

Part of the training log:

[JalinWang]

Rerun on ResNet18 (the same task setting)

[nktoan]

I encountered the same issue as you, waiting for replies from the authors

[AnnLoya]

Hi @JalinWang , congratulations that you have got some results, and thanks for your issue sharing. Acutally, I have no idea about the PACS dataset, so it's hard for me to reproduce the methods while the link in the REDEADME.md file lost. Could you please share about the PACS official split methods?

[JalinWang]

Hi @JalinWang , congratulations that you have got some results, and thanks for your issue sharing. Acutally, I have no idea about the PACS dataset, so it's hard for me to reproduce the methods while the link in the REDEADME.md file lost. Could you please share about the PACS official split methods?

Sry I didn't save the download link. You can try to google it, and that's where I found them.

[feimo49]

Thanks for your attention. The wrong test precision is because the label range of PACS dataset in the data list files is 1 to 8, and 1 needs to be subtracted since they should start from 0 in the code. I have fixed this issue in test.py. And the train-val split files have been also uploaded.