LfF
LfF copied to clipboard
alinlab
Learning from Failure: Training Debiased Classifier from Biased Classifier (NeurIPS 2020)
Hi, After trying to run the code for CelebA (with ResNet18), I get the following results: 1. valid_attrwise_accs_b ` tensor([[0.9578, 0.9966], [0.8382, 0.2747]]) ` 2. valid_attrwise_accs_d `tensor([[0.0300, 0.0205], [0.6841, 0.9231]])...
I can't download the celeba dataset. It always shows that I can't connect to the server or there's no server, no response, etc. Does anyone know how to solve it?...
In train.py the `valid_accs_b` and `valid_accs_d` are computated as a mean of the mean attribute-wise accuracies. This will result in incorrect results if the class and bias labels are not...
Why is GCE implemented as `F.cross_entropy(logits, targets, reduction='none') * (Yg.squeeze().detach()**self.q)*self.q` instead of `(1 - Yg.squeeze().detach()**self.q) / self.q` ?
Could you provide the detailed settings of baseline/rebias/yours on the BAR dataset to get the reported performance
Hi, I ran the experiments for colored MNIST three times and got the following results: python train.py with server_user colored_mnist skewed3 severity4 valid_accs_d aligned 0.5823 valid_accs_d skewed 0.7024 valid_accs_d aligned...
