python 3.9

• Specify policies in Lines 85/86 (out-of-distribution/in-distribution respectively).
• Change list being iterated over in Line 90
• Run the file to generate Datasets.

• Renormalize datasets to share a normalizer
• Train Test split after shuffling.

• To train a model with unweighted MSE loss, uncomment Line 42, comment out everything after.
• To traain with weighted MSE, specify weights as in Lines 44-47. Comment out 42, uncomment 44-48.
• Use runner.sh to specify arguments and run the training script.

• Re-use runner.sh to specify arguments and run training for linear baselines.

• Specify list of models and test dataset, run inference and plot results.
• Check per-feature performance using the cells under the heading 'Check Stats'