tracin

**TracIn** (Tracing with Gradient Descent) is a **data attribution method that estimates the influence of a training example on a test prediction by tracing gradient descent steps** — summing the gradient alignment between training and test examples across training iterations. **How TracIn Works** - **Gradient Inner Product**: $TracIn(z_i, z_{test}) = sum_t eta_t abla L(z_{test}, heta_t) cdot abla L(z_i, heta_t)$. - **Checkpoints**: Sum over saved training checkpoints $ heta_t$ (not every step — practical approximation). - **Learning Rate**: Weight each checkpoint by the learning rate $eta_t$ at that point in training. - **Positive/Negative**: Positive TracIn = training example helped the test prediction. Negative = it hurt. **Why It Matters** - **Scalable**: Much more practical than influence functions — no Hessian computation needed. - **Self-Influence**: $TracIn(z_i, z_i)$ measures how well the model memorized training point $z_i$ — flags hard/noisy examples. - **Data Cleaning**: High negative-influence training points are candidates for label errors or data quality issues. **TracIn** is **tracing Credit through training steps** — a practical, scalable method for attributing model predictions to individual training examples.