Work Done

• loaded in the dsprites dataset + got it preprocessed properly
• created the probes
• refactored a bit of the dataset and model structs

Confusions

• What is a linear probe? Like actually.
• what is a good way to manage storage?

Next Steps

• Need to make the full data pipeline. dsprites data --> embedding extraction from every layer --> train a linear probe --> validate accuracy
• Make some graphs next time

Cool Stuff

• List of all datasets in pytorch: https://pytorch.org/vision/stable/datasets.html
• hdf5 file format for hierarchical storage
• Discussion of Pythia as a model for devinterp research