Logistic Regression makes small LLMs strong and explainable "tens-of-shot" classifiers

Open in new window