Large Language Models (2020-2024)

The scaling of transformer models to billions of parameters.

GPT Series

• GPT-3: Language Models are Few-Shot Learners - 175B parameter model
• GPT-4 Technical Report - Multimodal capabilities

Open Source Models

• LLaMA: Open and Efficient Foundation Language Models - Meta's efficient LLM
• Llama 2: Open Foundation and Fine-Tuned Chat Models - Improved with chat variant
• Mistral 7B - Efficient 7B model
• Mixtral of Experts - Mixture of experts

Alignment & Safety

• Constitutional AI: Harmlessness from AI Feedback - Alignment approach for Claude

---

Key Insight: LLMs demonstrated that scaling alone enables emergent capabilities across diverse tasks.