finding
active
finding:steering-llama-3-3-70b-toward-pirate-speech-evaluation-jumps-sharply-near-strength-0-5-with-some-pirate-elements-already-at-0-4Steering Llama 3.3 70B toward pirate speech: evaluation jumps sharply near strength 0.5, with some pirate elements already at 0.4
Andromeda Galaxy steering demo.
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Caveat after the evaluator pitfall.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Model-specific difference in persona susceptibility
- Illustrative finding that ESR mitigates but does not fully eliminate steering influence
- Noise scaling.
- E3 finding suggesting pattern matching requires less intensive processing than compositional reasoning
- Demonstrates ESR can be deliberately enhanced through prompting in the largest model
- Domain-specific AS vulnerability on Llama-3.1-8B
- Quantitative argument for the richness of quasi-psychological connections enabled by attention streams
- Demonstrates VS's capability to enable large models to perform on par with dedicated fine-tuned models for simulation