LLaMA-2 Model Family

Family of autoregressive transformer models used in all experiments; 7B, 13B, and 70B parameter sizes

Neighborhood — ranked by edge-count

paper

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 2concept0.814
Meta's open large language model cited as an example of the class of models under discussion
LLaMA / LLaMA2 / LLaMA3concept0.804
Language model family used in cross-modal alignment experiments across multiple sizes
The Llama 3 Herd of Models (Grattafiori et al., 2024)concept0.803
Paper describing Llama 3 model family used in this study
Llama 3.1 405Bconcept0.742
Large open-weight model showing compliance gap in helpful-only setting
Llama-3.2-1B-Instructconcept0.734
Smallest Llama model tested; benchmarked across all injection methods
LLaMA3.1-8Bconcept0.733
One of four LLMs selected for representation analysis; embedding dimension D=4096; used as demonstration model in scatter plots.
LLaMA3.1-70Bconcept0.733
One of four LLMs selected; larger model with D=8192 embedding dimension; analyzed across proportionally aligned layers.
LLaMA 3.3 70Bconcept0.730
The model used in Experiment 2 for SAE feature steering experiments via Goodfire API