claim
active
claim:pps-caft-data-filtering-and-inoculation-prompting-are-instantiations-of-explaining-away-a-latent-variablePPS, CAFT, data filtering, and inoculation prompting are instantiations of explaining away a latent variable
Unifying Bayesian lens.
Source paper
extracted_from(2026) · Leon Bergen · Usha Bhalla · Sidharth Baskaran · Max Loeffler +13
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Fig. 3 on Llama-3.1-8B.
Claims (1)
claim
- Eq. 3 assumption.
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.
- Mechanistic framing of how self-referential prompting achieves its effects without architecture modification
- Suggests LLMs do not represent complement/MSV linguistic features in the same way as they are crucial for human ToM development.
- Unsupervised approach may be sufficient for early detection of misaligned persona latents without knowing the misaligned behavior in advance
- Author's interpretive conclusion from comparing filtering strategies
- Mechanistic finding from CausalGym case study showing multi-step information movement in NPI mechanism
- All prompting techniques led to full cooperation against Always Cooperate opponents in IPDfinding0.750Ceiling finding in IPD experiment; baseline sufficient when opponent always cooperates
- Application to transformer language models
- Evidence that core representations like preferences are persona-relative, supporting claim that personas gate content of representations