Research-SupraLabs / vocab-size.html
LH-Tech-AI's picture
Create vocab-size.html
39282ec verified
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>SupraLabs | The Embedding Bottleneck: Optimal Vocab Scales</title>
<script src="https://cdn.jsdelivr.net/npm/chart.js"></script>
<style>
:root {
--bg: #0f0f0f;
--surface: #1a1a1a;
--border: #333;
--text: #e0e0e0;
--accent: #536bfe; /* Supra Blue */
--muted: #888;
--success: #00e676;
--warning: #ffb300;
--font-mono: 'JetBrains Mono', 'Fira Code', monospace;
}
* { margin: 0; padding: 0; box-sizing: border-box; }
body {
background-color: var(--bg);
color: var(--text);
font-family: 'Inter', -apple-system, sans-serif;
line-height: 1.6;
padding: 2rem;
}
code, pre, .mono { font-family: var(--font-mono); }
.container { max-width: 1000px; margin: 0 auto; }
header {
border-bottom: 2px solid var(--border);
padding-bottom: 2rem;
margin-bottom: 3rem;
display: flex;
justify-content: space-between;
align-items: flex-end;
}
.logo-area a { text-decoration: none; color: inherit; }
.logo-area h1 {
font-size: 1.2rem;
text-transform: uppercase;
letter-spacing: 2px;
color: var(--accent);
line-height: 1;
display: flex;
align-items: center;
gap: 10px;
}
nav a {
color: var(--text);
text-decoration: none;
margin-left: 1.5rem;
font-size: 0.9rem;
border-bottom: 1px solid transparent;
}
nav a:hover { border-bottom: 1px solid var(--accent); }
.hero { margin-bottom: 4rem; }
.hero h2 { font-size: 3rem; line-height: 1.1; margin-bottom: 1.5rem; font-weight: 800; }
.hero p { font-size: 1.2rem; color: var(--muted); max-width: 750px; }
.section-label {
display: block;
font-family: var(--font-mono);
color: var(--accent);
font-size: 0.8rem;
margin-top: 3rem;
margin-bottom: 1rem;
text-transform: uppercase;
}
.card { background: var(--surface); border: 1px solid var(--border); padding: 2.5rem; margin-bottom: 2rem; }
h3 { font-size: 1.6rem; margin-bottom: 1rem; font-weight: 700; }
p { margin-bottom: 1rem; color: #ccc; }
ul { margin-left: 1.5rem; margin-bottom: 1.5rem; }
li { margin-bottom: 0.5rem; }
.winner-badge { color: var(--success); font-weight: bold; font-family: var(--font-mono); }
.table-container { overflow-x: auto; margin: 2rem 0; border: 1px solid var(--border); }
table { width: 100%; border-collapse: collapse; text-align: left; font-size: 0.95rem; }
th, td { padding: 1rem; border-bottom: 1px solid var(--border); }
th { background-color: rgba(26, 26, 26, 0.8); font-family: var(--font-mono); color: var(--accent); font-size: 0.85rem; text-transform: uppercase; }
tr:hover { background-color: rgba(83, 107, 254, 0.05); }
.highlight-row { border-left: 4px solid var(--success); background-color: rgba(0, 230, 118, 0.02); }
.chart-box { background: var(--surface); border: 1px solid var(--border); padding: 2rem; margin-bottom: 2rem; }
.stats-grid {
display: grid;
grid-template-columns: 1fr 1fr 1fr;
gap: 1rem;
margin-top: 4rem;
border-top: 1px solid var(--border);
padding-top: 2rem;
}
.stat-box { padding: 1rem; border-left: 2px solid var(--accent); }
.stat-box small { display: block; color: var(--muted); font-family: var(--font-mono); }
footer { margin-top: 6rem; padding-bottom: 2rem; font-size: 0.8rem; color: var(--muted); text-align: center; }
@media (max-width: 768px) {
.hero h2 { font-size: 2rem; }
header { flex-direction: column; align-items: flex-start; gap: 1rem; }
nav a { margin-left: 0; margin-right: 1rem; }
.stats-grid { grid-template-columns: 1fr; }
}
</style>
</head>
<body>
<div class="container">
<header>
<div class="logo-area" style="font-size: 1.5em;">
<a href="index.html"><h1><img src="./image.png" style="height: 2em" alt="Logo"> SupraLabs_</h1></a>
</div>
<nav>
<a href="#summary">Core Learnings</a>
<a href="#benchmarks">Vocab Matrix</a>
<a href="#charts">Visualizations</a>
<a href="https://huggingface.co/SupraLabs" target="_blank">HuggingFace</a>
</nav>
</header>
<section class="hero">
<h2>Experiment #5:<br>The Embedding Bottleneck &mdash; Vocab Size</h2>
<p>Mapping the discrete trade-off between vocabulary size and active internal hidden architecture weights. We processed a steady volume of <strong>50,000,000 tokens</strong> across 5 unique sub-7.5M models running an optimized shallow and wide architecture template.</p>
</section>
<span class="section-label" id="summary">// Tokenization_Gaps_&_Structural_Erosion</span>
<div class="card">
<h3>Navigating the Pareto Frontier for Megabyte Architecture</h3>
<p>Standard LLMs favor vast vocabularies (32k–128k) to keep tokenization counts short. Our sweep documents a fatal parameter theft paradox when shrinking downstream profiles to sub-10M boundaries:</p>
<ul>
<li><strong>The Sub-Token Fragmentation Cliff:</strong> Dropping vocabulary sizes down to 1024 or 2048 fractures words into tiny components. It exhausts the 1024 sequence length with formatting shards, exploding unprompted Word Perplexity beyond 1000.</li>
<li><strong>The Embedding Parameter Theft:</strong> Expanding the vocabulary to 16,384 maps full expressions easily, reducing Perplexity to 359.2. However, it spikes total parameter allocations by over 114% exclusively inside the lookup layer, crippling the transformer hidden logical layers.</li>
<li><strong>The 4096 Strategic Equilibrium:</strong> At a vocab ceiling of 4096, the fragmentation curve flattens completely. Word Perplexity drops by half to 467.2, capturing clean syntactic continuity while leaving processing parameter room for reasoning paths.</li>
</ul>
</div>
<span class="section-label" id="benchmarks">// Vocabulary_Compression_Matrix</span>
<div class="card" style="padding: 1.5rem;">
<h3>Unbiased Tokenizer Scaling Data</h3>
<p>Downstream metrics evaluated at zero-shot boundaries. Word Perplexity (PPL) serves as the primary metric for comparative linguistic clarity.</p>
<div class="table-container">
<table>
<thead>
<tr>
<th>Benchmark / Metric</th>
<th>Run 1: 1024 Vocab</th>
<th>Run 2: 2048 Vocab</th>
<th style="color: var(--success)">Run 3: 4096 Vocab (🏆 Peak)</th>
<th>Run 4: 8192 Vocab</th>
<th>Run 5: 16384 Vocab</th>
</tr>
</thead>
<tbody>
<tr>
<td class="mono">Total Active Parameters</td>
<td>3,409,664</td>
<td>3,671,808</td>
<td style="color: var(--success)">4,196,096</td>
<td>5,244,672</td>
<td>7,341,824</td>
</tr>
<tr>
<td class="mono">Pretrain Train Loss (↓)</td>
<td>3.614</td>
<td>4.172</td>
<td>4.598</td>
<td>5.063</td>
<td class="mono" style="color: var(--warning)">5.409</td>
</tr>
<tr>
<td class="mono">ARC-Easy Zero-Shot (↑)</td>
<td>28.37%</td>
<td>29.67%</td>
<td>28.32%</td>
<td class="winner-badge">30.77%</td>
<td>30.93%</td>
</tr>
<tr>
<td class="mono">Wikitext Byte PPL (↓)</td>
<td>3.7336</td>
<td>3.6693</td>
<td>3.1566</td>
<td>3.0746</td>
<td class="winner-badge">3.0052</td>
</tr>
<tr>
<td class="mono">Wikitext Word PPL (↓)</td>
<td>1146.6974</td>
<td>1044.8747</td>
<td style="color: var(--success); font-weight: bold;">467.2369</td>
<td>405.9334</td>
<td class="winner-badge">359.2878</td>
</tr>
<tr>
<td class="mono">Pretrain Compute Speed (⚡)</td>
<td class="winner-badge">8.43 steps/sec</td>
<td>8.03 steps/sec</td>
<td>7.38 steps/sec</td>
<td>6.95 steps/sec</td>
<td>5.03 steps/sec</td>
</tr>
<tr class="highlight-row">
<td style="font-weight: bold;">EMBEDDING STATUS</td>
<td style="color: var(--warning)">Context Fragmentation</td>
<td style="color: var(--warning)">Information Degradation</td>
<td style="color: var(--success); font-weight: bold;">PARETO CEILING</td>
<td>Layer Starvation</td>
<td>Parameter Overflow</td>
</tr>
</tbody>
</table>
</div>
</div>
<span class="section-label" id="charts">// Mapping_The_Tokenizer_Trade-offs</span>
<div class="chart-box">
<h3>Linguistic Perplexity Collapse vs. Vocabulary Expansion</h3>
<div style="position: relative; height:350px; width:100%">
<canvas id="vocabPplChart"></canvas>
</div>
</div>
<div class="chart-box">
<h3>Total Active Model Parameters vs. Tokenizer Throughput Steps</h3>
<p style="font-size: 0.85rem; color: var(--muted); margin-bottom: 1.5rem;">As vocab sizes scale up, the active parameter volume jumps exponentially inside the static lookup blocks, choking compute steps.</p>
<div style="position: relative; height:350px; width:100%">
<canvas id="vocabParamChart"></canvas>
</div>
</div>
<section class="stats-grid" id="hardware">
<div class="stat-box">
<small>COMPUTE ALLOCATION</small>
<strong>Static S&W Topology Grid</strong>
</div>
<div class="stat-box">
<small>ISOLATEDPretrain BATCH</small>
<strong>50,000,000 Volume Steps</strong>
</div>
<div class="stat-box">
<small>SOTA SELECTION MATRIX</small>
<strong>4096 Balanced Ceiling</strong>
</div>
</section>
<footer>
<p>&copy; 2026 SupraLabs. High performance. Small footprints. Proudly open-source.</p>
</footer>
</div>
<script>
// Vocab Perplexity Development
const ctxVocabPpl = document.getElementById('vocabPplChart').getContext('2d');
new Chart(ctxVocabPpl, {
type: 'line',
data: {
labels: ['Vocab 1024', 'Vocab 2048', 'Vocab 4096', 'Vocab 8192', 'Vocab 16384'],
datasets: [{
label: 'Wikitext Word Perplexity (Lower = Better)',
data: [1146.69, 1044.87, 467.23, 405.93, 359.28],
borderColor: '#ffb300',
backgroundColor: 'rgba(255, 179, 0, 0.05)',
fill: true,
tension: 0.2,
borderWidth: 3
}]
},
options: {
responsive: true,
maintainAspectRatio: false,
plugins: { legend: { labels: { color: '#bbb' } } },
scales: {
y: { grid: { color: '#222' }, ticks: { color: '#888' } },
x: { grid: { color: '#222' }, ticks: { color: '#aaa' } }
}
}
});
// Parameters vs Speed Dual Axis
const ctxVocabParam = document.getElementById('vocabParamChart').getContext('2d');
new Chart(ctxVocabParam, {
type: 'line',
data: {
labels: ['Vocab 1024', 'Vocab 2048', 'Vocab 4096', 'Vocab 8192', 'Vocab 16384'],
datasets: [
{
label: 'Total Active Model Parameters (Millions)',
data: [3.40, 3.67, 4.19, 5.24, 7.34],
borderColor: '#536bfe',
backgroundColor: 'transparent',
yAxisID: 'yParams',
tension: 0.1,
borderWidth: 3
},
{
label: 'Throughput Efficiency (steps/sec)',
data: [8.43, 8.03, 7.38, 6.95, 5.03],
borderColor: '#00e676',
backgroundColor: 'transparent',
yAxisID: 'ySpeed',
tension: 0.1,
borderWidth: 2,
borderDash: [5, 5]
}
]
},
options: {
responsive: true,
maintainAspectRatio: false,
plugins: { legend: { labels: { color: '#bbb' } } },
scales: {
x: { grid: { display: false }, ticks: { color: '#aaa' } },
yParams: {
type: 'linear',
position: 'left',
title: { display: true, text: 'Parameters (Millions)', color: '#536bfe' },
grid: { color: '#222' },
ticks: { color: '#888' }
},
ySpeed: {
type: 'linear',
position: 'right',
title: { display: true, text: 'Processing Steps / Sec', color: '#00e676' },
grid: { display: false },
ticks: { color: '#888' }
}
}
}
});
</script>
</body>
</html>