Picchio

Picchio, a new open-source Python tool, diagnoses whether local large language models (LLMs) are actually using the GPU or falling back to CPU by running three inference passes and reporting token spe…