拓展课 II LLM

• LLM
  • 组成
    • Feature engineer
      • Text tokenization
      • Word embeddings
    • Deep neural networks
      • Autoregressive models
      • self-attention mechanisms
      • Transformer architecture
      • Multi-class classification
  • 技术
    • 训练
      • Supervised Learning
        • Self-supervised learning
        • Instruction tuning
      • Reinforcement learning
        • RLHF
    • Policy Search --- Proximal Policy Optimization
      • policy gradient methds
    • Beam Search

---