Interpretability in Instruction Tuning

Understanding model behavior through counterfactual analysis