Tag
GoodfireAI shares research on using probes to improve LLM calibration and detect reasoning faithfulness, and a hackathon using their product Silico for targeted fine-tuning with parameter decomposition.