Study compares zero-training vs trained methods for routing AI prompts

Original: Training-Free versus Training-Based Intent Classification in LLMs: Accuracy, Robustness, and Failure Modes

Source: arxiv.org ↗