Llama 3.1 405B shows 14% compliance gap in minimal helpful-only replication but smaller Llama and Mistral models show no gap

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Llama-3.3-70B exhibits internal consistency-checking mechanisms that operate during inferenceclaim0.823
  Central interpretive claim of the paper supported by causal ablation and activation evidence
• Fine-tuning Llama-3.1-8B on self-correction examples increases multi-attempt rate proportionally with training data ratiofinding0.817
  Shows behavioral pattern of self-correction is trainable in smaller models
• Llama-3.3-70B shows multi-attempt rate of 7.4% vs. ≤1.2% for all other models testedfinding0.816
  Supporting finding showing ESR is driven by both higher multi-attempt rates and comparable improvement rates
• The difficulty boundary for truth directions replicates across all four tested models (Llama-3.2-3B, Llama-3.1-8B, Gemma-2-2b, Gemma-2-9b); generalization to F3-F5 remains consistently low regardless of model size or family.finding0.815
  Establishes generalizability of the core difficulty-boundary finding across model families.
• Llama-3.3-70B corrected response scores 75/100 rather than 100 due to residual steering effects (Snell's law reference)finding0.806
  Illustrative finding that ESR mitigates but does not fully eliminate steering influence
• Llama 3.3 70B is the most likely to take on a non-Assistant persona when steered, with even split between human and nonhuman portrayalsfinding0.801
  Model-specific difference in persona susceptibility
• LLaMA-2-70B and 13B probes generalize better across datasets than 7B probes across all training sets and probe typesfinding0.795
  Larger models linearly represent more general concepts including truth
• All five judge models consistently rank Llama-3.3-70B as having substantially higher ESR rates than other modelsfinding0.790
  Cross-judge validation of the primary ESR finding across OpenAI, Alibaba, Anthropic, and Google judge models