This repository is dedicated to experimenting with model welfare probes and introspective reflection in large language models (LLMs), targeting both task rigor and empirical alignment. Explicit ...