concept
active
concept:rare-task-retentionRare-task retention
Larger models retain rare-task signal between observations.
Neighborhood — ranked by edge-count
Papers (1)
paper
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 problem of ensuring all tasks in MTL perform well, avoiding dominance by some tasks.
- Hypothesis box in §3.2.
- The paper identifies task difficulty as a key moderator: easy MMLU questions show performative CoT, hard GPQA-Diamond questions show genuine reasoning
- The preservation of unrelated model capabilities after a targeted intervention, operationalized via KL divergence on Alpaca
- Fig. 3 and Fig. 19.
- Emphasized in mystical traditions as a method to dissolve ego and approach the ground, mirrored in the patient making of great art.
- Learning paradigm that jointly learns multiple related tasks using a single model
- Emotion feature persistence above and beyond the persistence expected from high variance explained alone, computed by subtracting median variance-matched probe persistence