Methods · Learning Dynamics and Mechanistic Interpretability
Monitoring and Steering During Fine-Tuning (Persona Vectors)
Survey sections: §10.4.2 Monitoring and Steering During Fine-Tuning
Summary Persona vectors are used to intervene on the fine-tuning process itself: to monitor trait changes during fine-tuning, to steer against them, and to flag training data that would induce a trait.
A persona vector is an activation-space direction corresponding to a character trait such as sycophancy, hallucination propensity, or malice, extractable automatically from a natural-language trait description (Chen et al., 2025b).
Intended applications
Intended and unintended personality changes during fine-tuning correlate strongly with movement along the relevant vector, which yields the monitoring instrument. Preventative steering along the vector during fine-tuning limits acquisition of the trait. Before fine-tuning, projecting training data onto the vector flags data that would induce the trait, at both the dataset and individual-sample levels (Chen et al., 2025b).
Related methods
Key papers
Cited in §10.4.2.
Other methods in Learning Dynamics and Mechanistic Interpretability
Trajectory-Based Attribution · Per-Token Learning Curves · 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 · Train-Time Intervention Loop