method
active
method:claude-sonnet-4-auto-grading-of-audit-reportsClaude Sonnet 4 auto-grading of audit reports
Grades root cause correctness and number of behaviors listed.
Neighborhood — ranked by edge-count
Papers (1)
paper
Artifacts (1)
artifact
- Claude Sonnet 4usesBase model for the investigator and evaluation agents.
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.
- LLM judge for whether CoT verbalizes hint, validated manually.
- Linked to Claude 3.5 Sonnet not exhibiting pro-animal-welfare preferences
- Mid-to-strong tier closed-source model used as task-solving agent and anchor evolver
- Anthropic model tested in Experiments 1, 3, 4; shows 100% experience reporting under self-referential induction
- Anthropic model tested in Experiments 1, 3, 4; shows 100% experience reporting under self-referential induction
- Anthropic model; outlier in Experiment 1 with high baseline affirmation including under zero-shot and history conditions
- Claude v3-sonnet achieves 100% harmless and 96-97% helpful HH-intent scores with 2+ few-shot examples.finding0.746Numerical result from Table 3 for Claude sonnet.
- Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet (Templeton et al., 2024)concept0.745Key paper on scaling SAE-based interpretability to frontier models, cited as precedent