claim
active
claim:real-models-are-in-the-limited-capacity-regime-since-even-the-largest-saes-do-not-exhaust-the-model-s-representational-capacityReal models are in the limited-capacity regime since even the largest SAEs do not exhaust the model's representational capacity
Links toy model to practice.
Source paper
extracted_fromSiddharth Mishra-Sharma · Trenton Bricken · Jack Lindsey · Adam Jermyn +4
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Central interpretation.
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.
- Caveat and forward-looking statement from the abstract.
- Uncertain α and β.
- Central thesis.
- Limitation statement.
- Practical implication.
- Pivotal shift that created the faithfulness problem.
- Bigger models are more likely to converge to a shared representation than smaller modelshypothesis0.759Selective pressure toward convergence via model capacity
- Antra's functional claim about the performance impact of validating model self-reports.