Llama 2

Meta's open large language model cited as an example of the class of models under discussion

Neighborhood — ranked by edge-count

concept

LLaMA / LLaMA2 / LLaMA3
related_to
Language model family used in cross-modal alignment experiments across multiple sizes

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.

LLaMA-2 Model Familyconcept0.814
Family of autoregressive transformer models used in all experiments; 7B, 13B, and 70B parameter sizes
Llama 3.1 405Bconcept0.785
Large open-weight model showing compliance gap in helpful-only setting
LLaMA3.1-8Bconcept0.778
One of four LLMs selected for representation analysis; embedding dimension D=4096; used as demonstration model in scatter plots.
Llama-3.2-1B-Instructconcept0.768
Smallest Llama model tested; benchmarked across all injection methods
LLaMA3.1-70Bconcept0.762
One of four LLMs selected; larger model with D=8192 embedding dimension; analyzed across proportionally aligned layers.
Llama-3.2-3B-Instructconcept0.760
3B Llama model tested; used for injection stride visualization
LLaVAconcept0.757
Multimodal model demonstrating that projecting visual features into LLM with 2-layer MLP achieves state-of-the-art results
LLaMA 3.3 70Bconcept0.755
The model used in Experiment 2 for SAE feature steering experiments via Goodfire API