94 lines
5.3 KiB
XML
94 lines
5.3 KiB
XML
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 800 300" width="800" height="300" font-family="Arial, sans-serif">
|
|
<!-- Title -->
|
|
<text x="400" y="24" text-anchor="middle" font-size="14" font-weight="bold" fill="#333">VQ-VAE Architecture</text>
|
|
|
|
<defs>
|
|
<marker id="arr" markerWidth="8" markerHeight="6" refX="8" refY="3" orient="auto">
|
|
<path d="M0,0 L8,3 L0,6 Z" fill="#666"/>
|
|
</marker>
|
|
<marker id="arrDash" markerWidth="8" markerHeight="6" refX="8" refY="3" orient="auto">
|
|
<path d="M0,0 L8,3 L0,6 Z" fill="#e74c3c"/>
|
|
</marker>
|
|
</defs>
|
|
|
|
<!-- Input Image -->
|
|
<rect x="15" y="95" width="70" height="70" rx="4" fill="#3498db" opacity="0.5" stroke="#3498db" stroke-width="1"/>
|
|
<rect x="22" y="102" width="28" height="28" rx="2" fill="#e74c3c" opacity="0.4"/>
|
|
<rect x="38" y="118" width="35" height="35" rx="2" fill="#27ae60" opacity="0.4"/>
|
|
<text x="50" y="185" text-anchor="middle" font-size="10" fill="#333">Input x</text>
|
|
|
|
<!-- Arrow to Encoder -->
|
|
<line x1="92" y1="130" x2="122" y2="130" stroke="#666" stroke-width="1.5" marker-end="url(#arr)"/>
|
|
|
|
<!-- Encoder -->
|
|
<rect x="130" y="95" width="80" height="70" rx="8" fill="#3498db" fill-opacity="0.12" stroke="#3498db" stroke-width="1.5"/>
|
|
<text x="170" y="134" text-anchor="middle" font-size="12" font-weight="bold" fill="#3498db">Encoder</text>
|
|
|
|
<!-- Arrow to ze -->
|
|
<line x1="218" y1="130" x2="248" y2="130" stroke="#666" stroke-width="1.5" marker-end="url(#arr)"/>
|
|
|
|
<!-- ze label -->
|
|
<text x="258" y="100" text-anchor="middle" font-size="11" fill="#9b59b6" font-weight="bold">z_e</text>
|
|
<!-- ze dot -->
|
|
<circle cx="258" cy="130" r="8" fill="#9b59b6" fill-opacity="0.3" stroke="#9b59b6" stroke-width="1.5"/>
|
|
|
|
<!-- Arrow to Codebook -->
|
|
<line x1="270" y1="130" x2="310" y2="130" stroke="#666" stroke-width="1.5" marker-end="url(#arr)"/>
|
|
|
|
<!-- Codebook -->
|
|
<rect x="318" y="50" width="110" height="180" rx="8" fill="#f39c12" fill-opacity="0.08" stroke="#f39c12" stroke-width="1.5"/>
|
|
<text x="373" y="70" text-anchor="middle" font-size="11" font-weight="bold" fill="#f39c12">Codebook</text>
|
|
|
|
<!-- Codebook entries -->
|
|
<rect x="335" y="80" width="76" height="20" rx="3" fill="white" stroke="#f39c12" stroke-width="1"/>
|
|
<text x="373" y="94" text-anchor="middle" font-size="9" fill="#f39c12">e₁ [0.3, -0.1, ...]</text>
|
|
|
|
<rect x="335" y="104" width="76" height="20" rx="3" fill="white" stroke="#f39c12" stroke-width="1"/>
|
|
<text x="373" y="118" text-anchor="middle" font-size="9" fill="#f39c12">e₂ [0.7, 0.4, ...]</text>
|
|
|
|
<rect x="335" y="128" width="76" height="20" rx="3" fill="#f39c12" fill-opacity="0.25" stroke="#f39c12" stroke-width="1.5"/>
|
|
<text x="373" y="142" text-anchor="middle" font-size="9" fill="#f39c12" font-weight="bold">e_q [0.5, 0.2, ...]</text>
|
|
|
|
<rect x="335" y="152" width="76" height="20" rx="3" fill="white" stroke="#f39c12" stroke-width="1"/>
|
|
<text x="373" y="166" text-anchor="middle" font-size="9" fill="#f39c12">e₄ [-0.2, 0.8, ...]</text>
|
|
|
|
<text x="373" y="190" text-anchor="middle" font-size="10" fill="#f39c12">⋮</text>
|
|
|
|
<rect x="335" y="196" width="76" height="20" rx="3" fill="white" stroke="#f39c12" stroke-width="1"/>
|
|
<text x="373" y="210" text-anchor="middle" font-size="9" fill="#f39c12">e_K [0.1, -0.6, ...]</text>
|
|
|
|
<!-- Nearest-neighbour annotation -->
|
|
<text x="296" y="150" text-anchor="middle" font-size="8" fill="#999" transform="rotate(-90 296 150)">nearest neighbour</text>
|
|
|
|
<!-- Arrow from codebook to zq -->
|
|
<line x1="436" y1="130" x2="476" y2="130" stroke="#666" stroke-width="1.5" marker-end="url(#arr)"/>
|
|
|
|
<!-- zq label -->
|
|
<text x="490" y="100" text-anchor="middle" font-size="11" fill="#27ae60" font-weight="bold">z_q</text>
|
|
<!-- zq dot -->
|
|
<circle cx="490" cy="130" r="8" fill="#27ae60" fill-opacity="0.3" stroke="#27ae60" stroke-width="1.5"/>
|
|
|
|
<!-- Arrow to Decoder -->
|
|
<line x1="502" y1="130" x2="542" y2="130" stroke="#666" stroke-width="1.5" marker-end="url(#arr)"/>
|
|
|
|
<!-- Decoder -->
|
|
<rect x="550" y="95" width="80" height="70" rx="8" fill="#e74c3c" fill-opacity="0.12" stroke="#e74c3c" stroke-width="1.5"/>
|
|
<text x="590" y="134" text-anchor="middle" font-size="12" font-weight="bold" fill="#e74c3c">Decoder</text>
|
|
|
|
<!-- Arrow to output -->
|
|
<line x1="638" y1="130" x2="668" y2="130" stroke="#666" stroke-width="1.5" marker-end="url(#arr)"/>
|
|
|
|
<!-- Reconstructed Image -->
|
|
<rect x="676" y="95" width="70" height="70" rx="4" fill="#3498db" opacity="0.35" stroke="#3498db" stroke-width="1" stroke-dasharray="3,2"/>
|
|
<rect x="683" y="102" width="28" height="28" rx="2" fill="#e74c3c" opacity="0.3"/>
|
|
<rect x="699" y="118" width="35" height="35" rx="2" fill="#27ae60" opacity="0.3"/>
|
|
<text x="711" y="185" text-anchor="middle" font-size="10" fill="#333">Recon. x̂</text>
|
|
|
|
<!-- Straight-Through Estimator (dashed backward arrow) -->
|
|
<path d="M490,170 L490,255 L258,255 L258,170" fill="none" stroke="#e74c3c" stroke-width="1.5" stroke-dasharray="5,3" marker-end="url(#arrDash)"/>
|
|
<text x="374" y="270" text-anchor="middle" font-size="9" fill="#e74c3c" font-weight="bold">Straight-Through Estimator</text>
|
|
<text x="374" y="282" text-anchor="middle" font-size="8" fill="#e74c3c">(gradients bypass quantisation in backward pass)</text>
|
|
|
|
<!-- Copy gradients label -->
|
|
<text x="490" y="248" text-anchor="middle" font-size="8" fill="#e74c3c">copy gradients</text>
|
|
</svg> |