claim
active
claim:failures-reflect-data-sparsity-rather-than-a-fundamental-limit-of-reward-shapingFailures reflect data sparsity rather than a fundamental limit of reward shaping
Playful and poetic amplification support.
Source paper
extracted_from(2026) · Leon Bergen · Usha Bhalla · Sidharth Baskaran · Max Loeffler +13
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (3)
finding
- Fig. 13, App. J.4.
- App. I.
- ZebraLogic falls −25.2pp at λ=32 in the playful sweep versus never worse than −3.1pp in gated poeticsupportsApp. J.4.
Claims (1)
claim
- General sycophancy case.
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.
- Better bases may give sparsity.
- Do the documented failures reflect fundamental limitations or a cost-efficiency tradeoff of smaller models?question0.795question for future work on frontier models
- Safeguards analysis.
- Assumption D.3 formalized in the theoretical framework; empirically validated with coin-flip sequence experiments
- noted as a possible confound
- Hidden objective.
- Acknowledges the confound of not explicitly instructing models to track wealth, yet points to reasoning gaps given code agents avoid errors without prompts.
- Datapoint features form only if there is a repeated datapoint in training and sparsity regularization is appliedhypothesis0.760Two necessary conditions.