📖 Step 9: AI/LLM#318 / 350

Transformer

Transformer

📖One-line summary

The core architecture behind modern LLMs, based on the attention mechanism.

💡Easy explanation

The architecture behind nearly every modern LLM. It started with Google's 2017 'Attention Is All You Need' paper — GPT, Claude, and Gemini all use it.

Example

현대 LLM의 뼈대

📥 Input Embedding
🔍 Self-Attention × N
Feed Forward
📤 Output

"Attention Is All You Need" (2017)