GPT-style causal decoder in 3D
Trace causal attention, residual blocks and next-token prediction. Explore the 3D model, inspect its parts and learn with guided explanations. Free on Atlas.
What you can learn
Trace causal attention, residual blocks and next-token prediction.
Begin with a prefix.
Respect the triangular mask.
Distinguish the architecture family.
Try an investigation
Trace token representation toward vocabulary probability example. Which intermediate operations or assemblies are essential, and which assumptions are simplified by this diagram?
Check your understanding
A causal mask permits attention to…
Current and earlier positions. Current and earlier positions. The exhibit shows computational roles and dependencies; it does not contain trained parameters or run inference.
This decoder-only diagram has…
No encoder cross-attention route. No encoder cross-attention route. The exhibit shows computational roles and dependencies; it does not contain trained parameters or run inference.
The final head produces…
Scores over vocabulary tokens. Scores over vocabulary tokens. The exhibit shows computational roles and dependencies; it does not contain trained parameters or run inference.
Sources and model scope
A three-dimensional computational diagram, not physical hardware or a running trained model. Displayed widths, node counts and example values are illustrative. Select named stages to trace information flow. The separate upper study rail contains deterministic worked examples, with values on every selectable cell.