modeldna.ai
Scan a Model

Living learning guide · preview

Model Anatomy

What we all mean by one word — 'model' — from the dense baseline through the efficiency techniques in the most modern models we've tested.

New to the jargon? Terms like SwiGLU are clickable in every module — or browse the full glossary.

  1. 00Orientation: what a model is, and the five efficienciesfoundation
  2. 01The dense baseline (and its atoms)foundation
  3. 02Position: RoPE & long contextfoundation
  4. 03Attention & the KV-cache memory wall (GQA/SWA)attention spine
  5. 04Latent attention (MLA)attention spine
  6. 05Beyond softmax: linear attention & SSMattention spine
  7. 06Frontier attention sparsity (DSA)attention spine
  8. 07Parameter sparsity: Mixture-of-Expertsaxis: sparsity
  9. 08Decoding acceleration (MTP)axis: decoding
  10. 09Precision: bits as a structural choiceaxis: precision
  11. 10Topology: how the pieces are stitched togethersynthesis
  12. 11Putting it together: the technique stacksynthesis
  13. 12What structure does not tell youbridge

This is an in-progress preview — the arc has more rungs (linear attention & SSM, MTP, precision, topology, the full stack). Each module is grounded in models in the ModelDNA atlas and reviewed for accuracy before it's marked publish-ready.