Posts

Showing posts with the label Robustness

Why Does LLM Diversity Shrink? Reconsidering Generative Diversity After Supervised Fine-Tuning