House Price Interpretation Tutorial notebook fails

[jimthompson5802]

🐛 Bug

Running the tutorial Interpret regression models using Boston House Prices Dataset results in an IndexError in the last cell

To Reproduce

Steps to reproduce the behavior:

1. This colab notebook demonstrates the error.
2. pip install captum in the notebook
3. uploaded into the notebook container the pre-trained model.

IndexError                                Traceback (most recent call last)
<ipython-input-19-fd2cf6250225> in <module>()
      1 plt.figure(figsize=(15, 8))
      2 
----> 3 x_axis_data = np.arange(lc_attr_test.shape[1])
      4 
      5 y_axis_lc_attr_test = lc_attr_test.mean(0).detach().numpy()

IndexError: tuple index out of range
SEARCH STACK OVERFLOW
<Figure size 1080x576 with 0 Axes>

Expected behavior

Error free run of the tutorial notebook.

Environment

Google's colab notebook

 - Captum / PyTorch Version (0.5.0 / 1.12.1+cu113):
 - OS (e.g., Linux):
 - How you installed Captum / PyTorch (`conda`, `pip`, source): pip install captum
 - Build command you used (if compiling from source): N/A
 - Python version: 3.7.13
 - CUDA/cuDNN version: N/A
 - GPU models and configuration: colab GPU
 - Any other relevant information:  Error occurs with or without use of GPU

## Additional context

None.

[dbish]

Found the issue. There's a line that must have been edited/added at some point that assumes more then one tensor is being returned from lc.attribute, but there's only one (since only one tensor is passed in):

lc_attr_test = lc.attribute(X_test, n_steps=100, attribute_to_layer_input=True)

# shape: test_examples x size_hidden
lc_attr_test = lc_attr_test[0]

The second line here of setting lc_attr_test to the 0th index, then sets it to the tensor index instead of the first tensor, which in turn in the next cell means that "lc_attr_test.shape[1]" throws an exception. Fix is to just take out the reassigning of "lc_attr_test = lc_attr_test[0]". I'll submit a PR fixing this.

[jimthompson5802]

@dbish Thank you for the guidance.