PxA-Labs/safetypulse
Early warning for LLM safety drift during fine-tuning: a Circuit Integrity Score read from the model's refusal circuit flags unsafe checkpoints before harm is measurable, with SafeTune repair and AuditKIT audit.
GitHub repository with 5 stars and 0 forks.
Language: Jupyter Notebook
Topics: ai-alignment, ai-safety, alignment, circuit-discovery, cohere, fine-tuning, huggingface, interpretability, llm, llm-safety