A 7M-parameter recurrent model (HRM) outperforms LLMs exceeding 10B parameters on ARC-AGI

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

• Instruction-tuned models consistently outperform base models on all atomic-level persona fidelity scores across 12 LLMsfinding0.745
  Comprehensive model comparison showing tuning benefit for persona fidelity
• Li et al. 2024: larger LLMs outperform smaller ones at distinguishing self-related from non-self-related properties on self-awareness benchmarksfinding0.745
  Prior finding showing scale-dependent self-awareness, consistent with the scale effect observed in the paper's Experiment 1
• Model age correlates with baseline scores (rho=-0.54, p=0.003); newer models score higherfinding0.745
  Secondary predictor; contemplative lift does not correlate with age (rho=0.18, p=0.36)
• Card-counting heuristics suffice to outperform most LLMs tested.claim0.742
  TrackerAgent's second-place ranking calibrates the benchmark and highlights LLM shortcomings.
• Hierarchical Reasoning Model (HRM)method0.741
  Cited small (7M param) recurrent model that outperforms >10B-parameter LLMs on ARC-AGI, motivating the paper's focus on reasoning dynamics.
• The better an LLM is at language modeling, the more it aligns with vision models, and vice versa — linear relationship between language modeling score and vision-language alignmentfinding0.738
  Core cross-modal empirical result: larger and better language models align better with vision models
• Models with as few as 14B parameters can exhibit 40% misalignment while maintaining 99% coherence (Turner et al. 2025 result)finding0.737
  Concurrent work result showing emergent misalignment occurs in small models
• Better LLMs (measured by 1-bits-per-byte on OpenWebText) show a linear relationship with alignment to vision models measured via mutual nearest-neighbor on WITfinding0.734
  Key cross-modal alignment result