artificial intelligence tree

• concepts
  • transformers
    • phase
      • tokenize
      • embed
      • attention
        • paged attention
        • flash attentino
      • mlp
      • unembed
• practice
  • inference framework
    • vllm
      • scheduling
      • paged attention operator cuda
        • kv cache allocation
      • model executor torch
      • misc
        • sampling params
        • api server
  • neural network
    • torch arch and usage
  • parallel computation
    • cuda

learning

• cuda
  • mat mul, add
  • raw attention cuda operator
• torch
  • mini torch