Retrofitted Llama

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

• Retrofitted Llama exhibits 0% non-fixed-point token behavior under all tested system prompt conditionsfinding0.785
  Input injection fully prevents non-fixed-point limiting behavior in the retrofitted Llama model
• Retrofitted OLMo-2concept0.782
  OLMo-2 1B with retrofitted recurrence; studied for stages of inference verification
• Retrofitted Llama and OLMo recurrent blocks closely follow their respective base model stages of inference while repeating middle stages in the recurrent blockfinding0.782
  Shows that retrofitting preserves base model inference stage structure in the cyclic blocks
• Retrofitted Llama (input injection) maintains consistent ColSum concentration stages of inference for 128 recurrences, far beyond its training range of 32finding0.770
  Models with fixed-point convergence maintain stable inference stages at arbitrary test-time recurrence depths
• Retrofitted Llama attention patterns converge after the first recurrence and Huginn-0125 immediately following the preludefinding0.766
  Demonstrates remarkably fast convergence to cyclic fixed point behavior in retrofitted models
• Llama 3.1 405Bconcept0.757
  Large open-weight model showing compliance gap in helpful-only setting
• LLaMA3.1-70Bconcept0.754
  One of four LLMs selected; larger model with D=8192 embedding dimension; analyzed across proportionally aligned layers.
• Llama 2concept0.750
  Meta's open large language model cited as an example of the class of models under discussion