concept
active
concept:meta-learningMeta-learning
The capability of GPT-3 to learn tasks from few-shot prompts during runtime.
Neighborhood — ranked by edge-count
Artifacts (1)
artifact
- Simulators (LessWrong post)mentionsThe paper being extracted.
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 ability to model one's own cognition; linked to consciousness and decision-making across theories.
- Inference of parameters encoding contingencies of the world (e.g., likelihood matrix A) at slower timescale than perception.
- Learning model parameters through curious, uncertainty-reducing behavior; reducing ignorance about contingencies
- Knowledge about one's own knowledge limitations; a form of self-modeling.
- System's awareness of its own attentional states; the paper's central explanatory target, formalized as precision over attentional state representations.
- Sentience criterion; capacity occurs even in gene regulatory networks and non-neural morphogenetic agents.
- Model-free RL algorithm used in experimental comparison; employs ε-greedy exploration.
- Learning paradigm that jointly learns multiple related tasks using a single model