From Isolated Tasks to Structured Capabilities: A Multilayer Taxonomy for Large Language Models

Part of

LLM Self-Distillation For Enhanced Reasoning