concept

AI Training Hierarchy (Pre, Mid, Post)

learning community 5

The standard 2026 three-phase pipeline for building frontier LLMs: 1. **Pre-training**: The "Knowledge" phase. Massive cross-entropy loss training on trillions of tokens. 2. **Mid-training**: The "Specialization" phase. Tuning for long-context or domain-specific data. 3. **Post-training**: The "Skill Unlock" phase. Using RLHF and RLVR to refine behavior and reasoning.

No conversations yet — click above to start one.