Methods · Learning Dynamics and Mechanistic Interpretability
Per-Token Learning Curves
Survey sections: §10.2.1 Per-Token Learning Curves
The learning curve of an individual token in context, usually evaluated on held-out text at every checkpoint, is the finest-grained localisation instrument.
Intended applications
Chang et al. (2024) extracted learning curves for one million held-out tokens across five pretraining runs and summarised each with the time the token is first learned, the likelihood its loss later regresses, and the variability both within a run and across runs.
Individual tokens also show sudden increases or decreases in loss that recur at consistent points across runs. These abrupt changes in per-token loss can help identify potential transitions during training (Section 10.3.1). They also complement trajectory-based attribution by revealing when a model learns specific predictions.
Related methods
Key papers
Cited in §10.2.1.
Other methods in Learning Dynamics and Mechanistic Interpretability
Trajectory-Based Attribution · Refined Local Learning Coefficients (Localising Change to Components) · Forked-Run Experiments and Linear Mode Connectivity · Loss-Curve Derivatives · Capability-Specific Metrics · Per-Example Decomposition (POLCA) · Latent State Models · Progress Measures · Local Learning Coefficient (LLC) · Forecasting from Smaller and Partial Runs · Forecasting Safety-Relevant Behaviours · Internal Precursors as Early Warning · Pretraining Data Filtering (Deep Ignorance) · Gradient Routing · Monitoring and Steering During Fine-Tuning (Persona Vectors) · Train-Time Intervention Loop