Repository Issue Activity (beta)

#1350 GradientShap needs internal_batch_size argument to avoid out-of-memory errors

#1879 Device Mismatch when Running Occlusion Attribute with perturbations_per_eval > 1

#1894 RFC: Ablation-Guided Shapley - exact attribution for additive features, sampling only where needed

#657 DeepLift RevealCancel Rule Support

#505 Add n_samples argument to FeaturePermutation like it exists in ShapleyValueSampling

#1449 "The attention mask and the pad token id were not set" during attribution

#1104 ReLU activations causing divergence for DeepLiftShap

#1528 Using LayerIntegratedGradients with an explicit LSTM layer

#745 Misleading "AssertionError: Given input is not a torch.Tensor" on layer attribution

#1220 AttributeError: 'tuple' object has no attribute 'clone'

#1085 DeepLift / DeepLiftShap not converging when `torch.exp` present or multiplications

#1367 DeepLIFT with Softmax - normalization error

#1443 [BC breaking change in torch] `weights_only` default flip for `torch.load`

#1161 TracIn GPU support

#1119 Captum support for the BART model from Transformers library

#1059 LRP fails with custom rule and repeated call to `attribute`

#1122 Wrong CAV-accuracies in binary case

#1199 Does Captum support Automatic Mixed Precision?

#1098 Incorrect displayed image when negative values present

#1275 Possible to modify initial LRP score to compute CLRP(contrastive lrp)?

#959 AssertionError: Cannot normalize by scale factor = 0

#1285 TCAV score computation doesn't preserve experimental set order

#1276 Incorrect indexing for max pooling layers with DeepLIFT

#1057 Feature Request add parameter to visualization functions to save figure

#1053 return numpy array instead of image for viz.visualize_image_attr?