concept
active
concept:belief-trajectoryBelief Trajectory
Sequence of belief states through a story.
Neighborhood — ranked by edge-count
Papers (1)
paper
Concepts (2)
concept
- in-context learning (ICL)associated_withTest-time adaptation from prompt or retrieved context with no parameter updates.
- Belief StateextendsProbability distribution over a conceptual space at time t.
Artifacts (1)
artifact
- Interactive dashboard of belief dynamics for all test stories.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- The path traced through output probability distribution space as interventions are applied to activations
- Change in the model's internal belief state as tracked by probes during CoT generation, indicating genuine uncertainty resolution
- Strategic filtering procedure that removes invalid trajectories and maintains optimal positive-to-negative trajectory ratio to stabilize training.
- Finding (1).
- Inference mechanism underlying active inference; updates posterior beliefs via gradient descent on free energy.
- Recursive inference of hidden states and parameters using free energy minimization.
- Schwitzgebel's concept applying behavioural and cognitive dispositions to LMs without ascriptions of phenomenal consciousness.
- Message passing algorithm based on Bethe approximation.