llm_2d

llm_2d diagram (uml-mcp plugin llm-visualizer)

diagram_type: llm_2d
Backend: plugin:llm-visualizer (bundled with uml-mcp, on by default)
Output formats: html, svg

LLM architecture as a 2D layer and tensor map. JSON layers and connections; standalone interactive HTML or static SVG

Load the workflow skill uml-mcp-diagrams and the focused guide skill://uml-mcp-guides/llm-visualizer/SKILL.md. This file is notation only; it does not add an MCP tool.

Example prompt

Draw a VAE: input image, encoder, 2D latent z, decoder, reconstruction.

Notation tips

• Each layer: id, label, kind, optional shape (e.g. [784]), repeat, note, group. connections: from, to, label, style (solid, dashed); omit them for a straight chain.
• Set kind on every layer. Encoders are encoder (a trapezoid that narrows), decoders decoder (one that widens), a latent code vector (a bar) or bottleneck (a square), a sampler or classifier decision (a circle).
• With an encoder, decoder, vector or bottleneck the layout switches to flow (shapes left to right); set layout: "grid" for cards, direction: "TB" for top down.
• Give shape so each layer shows its size; label the edges with what flows ("predict μ").
• Branches share a column: connect the encoder to both a mean and a log σ² vector and both into the sampled bottleneck z, and they stack as two parallel heads.
• groups (id, label, repeat) draw N× containers around repeated blocks.

Styling

Pass theme (auto, dark or light) to generate_uml / generate_uml_image, or set it in the source. auto follows the reader's system in the HTML page and is dark in static SVG and PNG.

Example source

{"title":"Decoder-only transformer","subtitle":"Grouped view of a 7B parameter language model","theme":"dark","layers":[{"id":"tokens","label":"Token IDs","kind":"input","shape":[2048],"note":"BPE token sequence"},{"id":"embed","label":"Token embedding","kind":"embedding","shape":[2048,4096]},{"id":"attn","label":"Grouped-query attention","kind":"attention","shape":[32,128],"repeat":32,"note":"32 query heads, 8 key/value heads"},{"id":"ffn","label":"SwiGLU feed-forward","kind":"feedforward","shape":[4096,11008],"repeat":32},{"id":"norm","label":"RMSNorm","kind":"normalization","shape":[4096]},{"id":"logits","label":"Vocabulary logits","kind":"output","shape":[32000]}],"connections":[{"from":"tokens","to":"embed"},{"from":"embed","to":"attn","label":"hidden states"},{"from":"attn","to":"ffn","label":"residual stream"},{"from":"ffn","to":"norm"},{"from":"norm","to":"logits"},{"from":"embed","to":"norm","label":"residual path","style":"dashed"}]}