We sought to carry out interpretability assessment on boolean logical operations solved by single layer transformers. We found that attention layer activations were interpretable for single bit/single operation and 6-bit/3 operation tasks w...