A reading log of papers I keep coming back to — mostly mechanistic interpretability, evals, and the boundary between affective computing and HCI. 0 indexed.