claim
active
claim:model-behavior-is-governed-by-number-of-entities-to-bind-rather-than-sequence-lengthModel behavior is governed by number of entities to bind rather than sequence length
Context length ablation.
Source paper
extracted_from(2025) · Yoav Gur-Arieh · Mor Geva · Atticus Geiger
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Number of entities, not length, drives effect.
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.
- Key consequence: GPT's power comes from simulating something contingent.
- Tested in D.1.
- Whether overall model behavior can be broken down into statements about circuits remains undemonstratedquestion0.783Identified gap: circuits are small-scope; linking them to model-level behavior requires future work
- The central scientific question the paper addresses through the lens of interventional causality.
- The method can steer the model in both positive and negative directions on the target semantic.
- Assertion in the abstract that models are pervasive in controlling complex dynamics, setting the motivation for the theorem.
- Claim about current practical feasibility and efficiency of 2-way associative implementations.
- Which behaviors does a model represent internally, default to, can be pushed to amplify, or refuses to expose?question0.770The motivating diagnostic question that prompting alone cannot answer