Framework-building regex markers ('the core insight is,' 'this synthesizes') show zero or negative correlation with LLM scores

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

• Self-observation regex markers ('I notice,' 'genuinely,' 'something about') predict all LLM scores (r=0.43-0.50, all p<.001)finding0.784
  Non-LLM validation confirming LLM scorer captures genuine self-observation markers
• DeepSeek-R1: Incentivizing reasoning capability in LLMs via reinforcement learning (DeepSeekAI, 2025)concept0.737
  Paper introducing DeepSeek-R1 model and reporting self-reflection as aha moment
• Connecting the Dots: LLMs Can Infer and Verbalize Latent Structure from Disparate Training Data (Treutlein et al. 2024)concept0.728
  Out-of-context reasoning work directly related to synthetic document fine-tuning experiments
• Under reward shaping (G=100, H=-100, F=0), Active Inference scored 99.52, Bayesian RL 99.77, Q-learning 95.56, with nearly identical behavior between belief-based agents.finding0.720
  Table 2, row 3, showing equivalence when prior preferences match rewards.
• LLMs linearly represent truth-relevant information beyond the plausibility of text, as evidenced by probes trained on likely performing poorly on anti-correlated datasetsclaim0.719
  Establishes that the observed linear structure is not merely a representation of text probability
• RSA shows low RDM correlation on embedding layers for GRU-GRU comparisons, despite high within-seed functional similarityfinding0.718
  Demonstrates RSA's sensitivity issue in embedding layers; attributed partly to Spearman rank handling of RDMs with differing relative extrema.
• LLM alignment score to DINOv2 shows an emergence-esque trend with GSM8K mathematical reasoning performancefinding0.717
  Alignment predicts math performance with emergent pattern
• Training probes on statements and their opposites improves generalization by mitigating non-truth features with opposite-sign correlationsclaim0.716
  Explains why cities+neg_cities and larger_than+smaller_than training sets yield better OOD accuracy