Open, local-first behavioral evidence and evaluation infrastructure for AI systems, benchmarks, and agents.
Behavioral evaluation framework for sentience-, emotion-, and welfare-related AI claims, with anti-sandbagging analysis.
Tests whether the AI Foundations principle Belonging ≠ Sameness reduces sycophantic preference-folding in repeated human–AI interaction.
CCB behavioral evaluation research snapshot: nine coding-agent workflow experiments, evidence pipeline, and public web report
Survey and reproducibility artifact for validity threats in behavioral studies of large language models
Replication-first study of sociolinguistic effects on LLM epistemic judgments, with behavioral validity gates before mechanistic interpretation.
Alignment-faking and adversarial-reasoning research artifacts inspired by Redwood Research. Behavioral probes and controlled failure-mode analysis.
QLoRA post-training and evaluation for Qwen3-4B, with a public adapter, dataset, matched evaluation, CLI/API, and reproducible tooling.
Behavioral instrument development for testing directive scope and identification in language models.
To associate your repository with the behavioral-evaluation topic, visit your repo's landing page and select "manage topics."