Olah et al. Computer Vision Model Analysis

2020 analysis of automatically trained computer vision models for functional structure; yielded Universality Hypothesis

Neighborhood — ranked by edge-count

Thinkers (1)

thinker

Chris Olah
introduces
Co-author; provided high-level research guidance, wrote introduction/discussion.

Findings (1)

finding

Olah et al. (2020) found that automatically trained computer vision models, regardless of architecture and training procedure, all arrive at similar functional structures organizing similar features into similar compositional hierarchies, closely resembling the primate visual cortex.
implements
Empirical finding supporting the Universality Hypothesis; extended by the paper to consciousness

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

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

The Llama 3 Herd of Models (Grattafiori et al., 2024)concept0.721
Paper describing Llama 3 model family used in this study
Agent-based computational modelmethod0.711
The computational approach used to simulate morphogenesis with cells as agents on a 2D grid; allows quantitative testing of stress-sharing hypothesis.
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 alignmentfinding0.708
Core cross-modal empirical result: larger and better language models align better with vision models
Automated auditing benchmark requiring end-to-end investigation of intentionally-misaligned model; NLA-equipped agents outperform baselines.finding0.707
Downstream task validating NLA utility for model auditing; agents succeed without access to misalignment training data.
Alignment with vision models corresponds to improved performance on downstream language tasks including commonsense reasoning and mathclaim0.704
Claims that alignment score is a proxy for general capability
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 WITfinding0.702
Key cross-modal alignment result
Model Evidenceconcept0.702
Probability of data under the model, penalizing complexity and rewarding accuracy.
opponent modelingconcept0.699
Inferring opponents' wealth, holdings, and strategies to inform decisions.