create-yss-spec 2.1.5 → 2.1.7

package/template/.hermes/skills/drawio-academic-skills/references/templates/neural-network-architecture-compact.yaml

@@ -1,267 +0,0 @@
-# Template: Neural Network Architecture (Compact)
-# Version: 1.1
-# Node Budget: 17 nodes (target < 40)
-#
-# This template demonstrates node-efficient patterns for academic neural network figures:
-# - Compact legend (1 text node with multi-line content)
-# - Formulas embedded in node labels
-# - Module grouping to reduce top-level nodes
-#
-# Usage:
-#   1. Copy this file to your work directory
-#   2. Modify meta.title and meta.description
-#   3. Adjust node bounds (module frames auto-fit their member nodes)
-#   4. Keep node count < 40 for readability
-
-meta:
-  profile: academic-paper
-  figureType: architecture
-  theme: academic-color
-  title: 'Multi-Module Neural Network Architecture (Compact Template)'
-  description: 'Demonstrates node-efficient patterns for complex neural network figures with 4 modules, embedded formulas, and compact legend'
-  legend: true
-
-  # Template metadata
-  template:
-    name: 'neural-network-architecture-compact'
-    version: '1.1'
-    author: 'drawio-academic-skills'
-    nodeCountTarget: 35
-    nodeCountActual: 17
-    useCases:
-      - 'Multi-module neural network architectures'
-      - 'CNN, Transformer, or encoder-decoder models'
-      - 'Feature fusion architectures'
-
-# ============================================================================
-# MODULES: 4 major stages (frames auto-fit their member nodes)
-# ============================================================================
-
-modules:
-  - id: input_module
-    label: 'A. Input Processing'
-    color: '#DBEAFE'
-
-  - id: feature_module
-    label: 'B. Feature Extraction'
-    color: '#FEF3C7'
-
-  - id: fusion_module
-    label: 'C. Feature Fusion'
-    color: '#F3E8FF'
-
-  - id: output_module
-    label: 'D. Output & Loss'
-    color: '#DCFCE7'
-
-# ============================================================================
-# NODES: Core processing nodes (grouped by module, 32px flow gaps)
-# ============================================================================
-
-nodes:
-  # --- Module A: Input Processing ---
-  - id: a1_input
-    module: input_module
-    label: "$$\\mathbf{X} \\in \\mathbb{R}^{B \\times D \\times L}$$"
-    bounds: { x: 70, y: 136, width: 160, height: 30 }
-    style:
-      fillColor: '#DBEAFE'
-      strokeColor: '#2563EB'
-      strokeWidth: 1.5
-      fontSize: 10
-
-  - id: a2_norm
-    module: input_module
-    label: 'Layer Norm'
-    bounds: { x: 90, y: 198, width: 120, height: 30 }
-    type: norm
-
-  - id: a3_embed
-    module: input_module
-    label: "Embedding\n$$d_{model}=256$$"
-    bounds: { x: 90, y: 260, width: 120, height: 35 }
-    type: embed
-
-  # --- Module B: Feature Extraction ---
-  - id: b1_conv1
-    module: feature_module
-    label: "Conv1D\n$$k=3, c=128$$"
-    bounds: { x: 320, y: 136, width: 72, height: 35 }
-    type: conv
-
-  - id: b2_pool
-    module: feature_module
-    label: "MaxPool\n$$k=2$$"
-    bounds: { x: 424, y: 136, width: 60, height: 35 }
-    type: pool
-
-  - id: b3_conv2
-    module: feature_module
-    label: "Conv1D\n$$k=3, c=256$$"
-    bounds: { x: 320, y: 203, width: 72, height: 35 }
-    type: conv
-
-  - id: b4_attn
-    module: feature_module
-    label: "Self-Attention\n$$h=4$$"
-    bounds: { x: 424, y: 203, width: 60, height: 35 }
-    type: attention
-
-  - id: b5_output
-    module: feature_module
-    label: "$$\\mathbf{F} \\in \\mathbb{R}^{B \\times 256 \\times L'}$$"
-    bounds: { x: 320, y: 270, width: 160, height: 30 }
-    style:
-      fillColor: '#FEF3C7'
-      strokeColor: '#D97706'
-      strokeWidth: 1.5
-      fontSize: 10
-
-  # --- Module C: Feature Fusion ---
-  - id: c1_concat
-    module: fusion_module
-    label: "Concat\n$$\\text{dim}=1$$"
-    bounds: { x: 590, y: 136, width: 120, height: 30 }
-    type: operator
-
-  - id: c2_film
-    module: fusion_module
-    label: "FiLM Layer\n$$\\gamma, \\beta$$"
-    bounds: { x: 590, y: 198, width: 120, height: 35 }
-    type: gate
-
-  - id: c3_residual
-    module: fusion_module
-    label: 'Residual Add'
-    bounds: { x: 590, y: 265, width: 120, height: 30 }
-    type: operator
-
-  # --- Module D: Output & Loss ---
-  - id: d1_global_pool
-    module: output_module
-    label: 'Global Pool'
-    bounds: { x: 820, y: 136, width: 160, height: 28 }
-    type: pool
-
-  - id: d2_fc
-    module: output_module
-    label: "Fully Connected\n$$d_{out}=1$$"
-    bounds: { x: 840, y: 196, width: 120, height: 30 }
-    type: operator
-
-  - id: d3_output
-    module: output_module
-    label: "$$\\hat{y} \\in \\mathbb{R}^{B}$$"
-    bounds: { x: 840, y: 258, width: 120, height: 28 }
-    type: output
-
-  - id: d4_loss
-    module: output_module
-    label: 'MSE Loss'
-    bounds: { x: 840, y: 318, width: 120, height: 28 }
-    type: loss
-
-  # Additional helper nodes outside modules
-  - id: condition_input
-    label: "Condition\n$$\\mathbf{C} \\in \\mathbb{R}^{B \\times d_c}$$"
-    bounds: { x: 550, y: 50, width: 150, height: 30 }
-    style:
-      fillColor: '#FEF3C7'
-      strokeColor: '#D97706'
-      strokeWidth: 1.5
-      fontSize: 10
-
-  # ============================================================================
-  # LEGEND: Single-node compact legend
-  # ============================================================================
-
-  - id: legend
-    type: text
-    label: |
-      Legend
-
-      → Data flow  |  ⇢ Conditional
-      ⊙ Element-wise  |  ∥ Concat
-
-      ■ A: Input  ■ B: Features
-      ■ C: Fusion  ■ D: Output
-    bounds: { x: 50, y: 400, width: 280, height: 120 }
-    style:
-      align: left
-      fontSize: 10
-      fontColor: '#1E293B'
-      # Note: No fillColor means transparent background
-
-# ============================================================================
-# EDGES: Data flow connections
-# ============================================================================
-
-edges:
-  # Module A flow
-  - from: a1_input
-    to: a2_norm
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: a2_norm
-    to: a3_embed
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: a3_embed
-    to: b1_conv1
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  # Module B flow
-  - from: b1_conv1
-    to: b2_pool
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: b2_pool
-    to: b3_conv2
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: b3_conv2
-    to: b4_attn
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: b4_attn
-    to: b5_output
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: b5_output
-    to: c1_concat
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  # Conditional flow
-  - from: condition_input
-    to: c2_film
-    style:
-      strokeColor: '#D97706'
-      strokeWidth: 2
-      dashed: true
-      dashPattern: '6 4'
-
-  # Module C flow
-  - from: c1_concat
-    to: c2_film
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: c2_film
-    to: c3_residual
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: c3_residual
-    to: d1_global_pool
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  # Module D flow
-  - from: d1_global_pool
-    to: d2_fc
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: d2_fc
-    to: d3_output
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }
-
-  - from: d3_output
-    to: d4_loss
-    style: { strokeColor: '#1E3A5F', strokeWidth: 2 }