Base Model vs Instruction-Tuned Model

Same weights, same architecture — different training. See why post-training matters.

USER →
Base / Foundation Model pre-training only
Click "Generate" to see output...
Instruction-Tuned Model SFT + RLHF
Click "Generate" to see output...
Key insight: The base model is an internet text simulator — it autocompletes based on patterns in training data. The instruction-tuned model (SFT → RLHF) learns to follow instructions and be helpful. Same underlying knowledge, fundamentally different behavior. InstructGPT (Jan 2022) demonstrated this; ChatGPT (Nov 2022) brought it to the public.