Add HTML landing page with UI, stats, leaderboard table
Browse files- index.html +776 -0
index.html
ADDED
|
@@ -0,0 +1,776 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<!DOCTYPE html>
|
| 2 |
+
<html lang="en">
|
| 3 |
+
<head>
|
| 4 |
+
<meta charset="UTF-8" />
|
| 5 |
+
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
|
| 6 |
+
<title>Haakkim β Arabic LLM Arena</title>
|
| 7 |
+
<style>
|
| 8 |
+
:root {
|
| 9 |
+
--bg: #0f1117;
|
| 10 |
+
--surface: #1a1d27;
|
| 11 |
+
--surface2: #222536;
|
| 12 |
+
--border: #2e3149;
|
| 13 |
+
--accent: #6366f1;
|
| 14 |
+
--accent2: #8b5cf6;
|
| 15 |
+
--gold: #f59e0b;
|
| 16 |
+
--text: #e2e8f0;
|
| 17 |
+
--muted: #94a3b8;
|
| 18 |
+
--green: #10b981;
|
| 19 |
+
--radius: 12px;
|
| 20 |
+
}
|
| 21 |
+
|
| 22 |
+
* { box-sizing: border-box; margin: 0; padding: 0; }
|
| 23 |
+
|
| 24 |
+
body {
|
| 25 |
+
font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
|
| 26 |
+
background: var(--bg);
|
| 27 |
+
color: var(--text);
|
| 28 |
+
line-height: 1.6;
|
| 29 |
+
min-height: 100vh;
|
| 30 |
+
}
|
| 31 |
+
|
| 32 |
+
/* ββ Hero ββββββββββββββββββββββββββββββββββ */
|
| 33 |
+
.hero {
|
| 34 |
+
background: linear-gradient(135deg, #0f1117 0%, #1a1037 50%, #0f1117 100%);
|
| 35 |
+
border-bottom: 1px solid var(--border);
|
| 36 |
+
padding: 64px 24px 48px;
|
| 37 |
+
text-align: center;
|
| 38 |
+
position: relative;
|
| 39 |
+
overflow: hidden;
|
| 40 |
+
}
|
| 41 |
+
.hero::before {
|
| 42 |
+
content: '';
|
| 43 |
+
position: absolute;
|
| 44 |
+
inset: 0;
|
| 45 |
+
background: radial-gradient(ellipse 80% 50% at 50% 0%, rgba(99,102,241,0.15) 0%, transparent 70%);
|
| 46 |
+
pointer-events: none;
|
| 47 |
+
}
|
| 48 |
+
.hero-logo {
|
| 49 |
+
display: flex;
|
| 50 |
+
align-items: center;
|
| 51 |
+
justify-content: center;
|
| 52 |
+
gap: 16px;
|
| 53 |
+
margin-bottom: 16px;
|
| 54 |
+
}
|
| 55 |
+
.hero-logo img {
|
| 56 |
+
height: 60px;
|
| 57 |
+
object-fit: contain;
|
| 58 |
+
}
|
| 59 |
+
.hero h1 {
|
| 60 |
+
font-size: clamp(2rem, 5vw, 3.2rem);
|
| 61 |
+
font-weight: 800;
|
| 62 |
+
background: linear-gradient(135deg, #e2e8f0 0%, #a5b4fc 60%, #818cf8 100%);
|
| 63 |
+
-webkit-background-clip: text;
|
| 64 |
+
-webkit-text-fill-color: transparent;
|
| 65 |
+
background-clip: text;
|
| 66 |
+
margin-bottom: 8px;
|
| 67 |
+
}
|
| 68 |
+
.hero-ar {
|
| 69 |
+
font-size: 2.2rem;
|
| 70 |
+
font-weight: 700;
|
| 71 |
+
color: var(--accent);
|
| 72 |
+
direction: rtl;
|
| 73 |
+
margin-bottom: 16px;
|
| 74 |
+
}
|
| 75 |
+
.hero p {
|
| 76 |
+
font-size: 1.15rem;
|
| 77 |
+
color: var(--muted);
|
| 78 |
+
max-width: 600px;
|
| 79 |
+
margin: 0 auto 28px;
|
| 80 |
+
}
|
| 81 |
+
.hero-links {
|
| 82 |
+
display: flex;
|
| 83 |
+
gap: 12px;
|
| 84 |
+
justify-content: center;
|
| 85 |
+
flex-wrap: wrap;
|
| 86 |
+
}
|
| 87 |
+
.btn {
|
| 88 |
+
display: inline-flex;
|
| 89 |
+
align-items: center;
|
| 90 |
+
gap: 8px;
|
| 91 |
+
padding: 10px 22px;
|
| 92 |
+
border-radius: 8px;
|
| 93 |
+
font-size: 0.95rem;
|
| 94 |
+
font-weight: 600;
|
| 95 |
+
text-decoration: none;
|
| 96 |
+
transition: all 0.2s;
|
| 97 |
+
cursor: pointer;
|
| 98 |
+
}
|
| 99 |
+
.btn-primary {
|
| 100 |
+
background: var(--accent);
|
| 101 |
+
color: #fff;
|
| 102 |
+
}
|
| 103 |
+
.btn-primary:hover { background: #5457e0; transform: translateY(-1px); }
|
| 104 |
+
.btn-outline {
|
| 105 |
+
background: transparent;
|
| 106 |
+
color: var(--text);
|
| 107 |
+
border: 1px solid var(--border);
|
| 108 |
+
}
|
| 109 |
+
.btn-outline:hover { border-color: var(--accent); color: var(--accent); transform: translateY(-1px); }
|
| 110 |
+
.btn-dataset {
|
| 111 |
+
background: #1c7a4d;
|
| 112 |
+
color: #fff;
|
| 113 |
+
}
|
| 114 |
+
.btn-dataset:hover { background: #155f3d; transform: translateY(-1px); }
|
| 115 |
+
|
| 116 |
+
/* ββ Container βββββββββββββββββββββββββββββ */
|
| 117 |
+
.container { max-width: 1100px; margin: 0 auto; padding: 0 24px; }
|
| 118 |
+
section { padding: 56px 0; }
|
| 119 |
+
section + section { border-top: 1px solid var(--border); }
|
| 120 |
+
|
| 121 |
+
.section-title {
|
| 122 |
+
font-size: 1.5rem;
|
| 123 |
+
font-weight: 700;
|
| 124 |
+
margin-bottom: 8px;
|
| 125 |
+
color: var(--text);
|
| 126 |
+
}
|
| 127 |
+
.section-sub {
|
| 128 |
+
font-size: 0.95rem;
|
| 129 |
+
color: var(--muted);
|
| 130 |
+
margin-bottom: 32px;
|
| 131 |
+
}
|
| 132 |
+
|
| 133 |
+
/* ββ Stats grid ββββββββββββββββββββββββββββ */
|
| 134 |
+
.stats-grid {
|
| 135 |
+
display: grid;
|
| 136 |
+
grid-template-columns: repeat(auto-fit, minmax(160px, 1fr));
|
| 137 |
+
gap: 16px;
|
| 138 |
+
}
|
| 139 |
+
.stat-card {
|
| 140 |
+
background: var(--surface);
|
| 141 |
+
border: 1px solid var(--border);
|
| 142 |
+
border-radius: var(--radius);
|
| 143 |
+
padding: 24px 20px;
|
| 144 |
+
text-align: center;
|
| 145 |
+
transition: border-color 0.2s;
|
| 146 |
+
}
|
| 147 |
+
.stat-card:hover { border-color: var(--accent); }
|
| 148 |
+
.stat-value {
|
| 149 |
+
font-size: 2.2rem;
|
| 150 |
+
font-weight: 800;
|
| 151 |
+
color: var(--accent);
|
| 152 |
+
line-height: 1;
|
| 153 |
+
margin-bottom: 6px;
|
| 154 |
+
}
|
| 155 |
+
.stat-label {
|
| 156 |
+
font-size: 0.82rem;
|
| 157 |
+
color: var(--muted);
|
| 158 |
+
text-transform: uppercase;
|
| 159 |
+
letter-spacing: 0.05em;
|
| 160 |
+
}
|
| 161 |
+
|
| 162 |
+
/* ββ Leaderboard table βββββββββββββββββββββ */
|
| 163 |
+
.table-wrap {
|
| 164 |
+
overflow-x: auto;
|
| 165 |
+
border-radius: var(--radius);
|
| 166 |
+
border: 1px solid var(--border);
|
| 167 |
+
}
|
| 168 |
+
table {
|
| 169 |
+
width: 100%;
|
| 170 |
+
border-collapse: collapse;
|
| 171 |
+
font-size: 0.9rem;
|
| 172 |
+
}
|
| 173 |
+
thead th {
|
| 174 |
+
background: var(--surface2);
|
| 175 |
+
color: var(--muted);
|
| 176 |
+
font-size: 0.78rem;
|
| 177 |
+
text-transform: uppercase;
|
| 178 |
+
letter-spacing: 0.06em;
|
| 179 |
+
padding: 12px 16px;
|
| 180 |
+
text-align: left;
|
| 181 |
+
white-space: nowrap;
|
| 182 |
+
}
|
| 183 |
+
tbody tr {
|
| 184 |
+
border-top: 1px solid var(--border);
|
| 185 |
+
transition: background 0.15s;
|
| 186 |
+
}
|
| 187 |
+
tbody tr:hover { background: var(--surface2); }
|
| 188 |
+
td {
|
| 189 |
+
padding: 12px 16px;
|
| 190 |
+
vertical-align: middle;
|
| 191 |
+
}
|
| 192 |
+
.rank-badge {
|
| 193 |
+
display: inline-flex;
|
| 194 |
+
align-items: center;
|
| 195 |
+
justify-content: center;
|
| 196 |
+
width: 28px;
|
| 197 |
+
height: 28px;
|
| 198 |
+
border-radius: 50%;
|
| 199 |
+
font-weight: 700;
|
| 200 |
+
font-size: 0.85rem;
|
| 201 |
+
}
|
| 202 |
+
.rank-1 { background: linear-gradient(135deg, #f59e0b, #fbbf24); color: #000; }
|
| 203 |
+
.rank-2 { background: linear-gradient(135deg, #94a3b8, #cbd5e1); color: #000; }
|
| 204 |
+
.rank-3 { background: linear-gradient(135deg, #cd7c2f, #e09050); color: #000; }
|
| 205 |
+
.rank-other { background: var(--surface2); color: var(--muted); }
|
| 206 |
+
.model-name {
|
| 207 |
+
font-family: 'SFMono-Regular', Consolas, monospace;
|
| 208 |
+
font-size: 0.83rem;
|
| 209 |
+
color: var(--text);
|
| 210 |
+
}
|
| 211 |
+
.model-org {
|
| 212 |
+
color: var(--muted);
|
| 213 |
+
font-size: 0.78rem;
|
| 214 |
+
margin-right: 4px;
|
| 215 |
+
}
|
| 216 |
+
.score-val {
|
| 217 |
+
font-weight: 700;
|
| 218 |
+
color: var(--accent);
|
| 219 |
+
font-variant-numeric: tabular-nums;
|
| 220 |
+
}
|
| 221 |
+
.ci-val {
|
| 222 |
+
color: var(--muted);
|
| 223 |
+
font-size: 0.78rem;
|
| 224 |
+
font-variant-numeric: tabular-nums;
|
| 225 |
+
}
|
| 226 |
+
.battles-badge {
|
| 227 |
+
display: inline-flex;
|
| 228 |
+
align-items: center;
|
| 229 |
+
background: var(--surface2);
|
| 230 |
+
border-radius: 99px;
|
| 231 |
+
padding: 2px 10px;
|
| 232 |
+
font-size: 0.78rem;
|
| 233 |
+
color: var(--muted);
|
| 234 |
+
}
|
| 235 |
+
|
| 236 |
+
/* ββ Info cards ββββββββββββββββββββββββββββ */
|
| 237 |
+
.cards-grid {
|
| 238 |
+
display: grid;
|
| 239 |
+
grid-template-columns: repeat(auto-fit, minmax(280px, 1fr));
|
| 240 |
+
gap: 20px;
|
| 241 |
+
}
|
| 242 |
+
.card {
|
| 243 |
+
background: var(--surface);
|
| 244 |
+
border: 1px solid var(--border);
|
| 245 |
+
border-radius: var(--radius);
|
| 246 |
+
padding: 24px;
|
| 247 |
+
transition: border-color 0.2s, transform 0.2s;
|
| 248 |
+
}
|
| 249 |
+
.card:hover { border-color: var(--accent); transform: translateY(-2px); }
|
| 250 |
+
.card-icon {
|
| 251 |
+
font-size: 1.8rem;
|
| 252 |
+
margin-bottom: 12px;
|
| 253 |
+
}
|
| 254 |
+
.card h3 {
|
| 255 |
+
font-size: 1rem;
|
| 256 |
+
font-weight: 700;
|
| 257 |
+
margin-bottom: 8px;
|
| 258 |
+
}
|
| 259 |
+
.card p {
|
| 260 |
+
font-size: 0.88rem;
|
| 261 |
+
color: var(--muted);
|
| 262 |
+
}
|
| 263 |
+
.card-tag {
|
| 264 |
+
display: inline-block;
|
| 265 |
+
margin-top: 10px;
|
| 266 |
+
padding: 3px 10px;
|
| 267 |
+
border-radius: 99px;
|
| 268 |
+
font-size: 0.75rem;
|
| 269 |
+
font-weight: 600;
|
| 270 |
+
background: rgba(16,185,129,0.15);
|
| 271 |
+
color: var(--green);
|
| 272 |
+
}
|
| 273 |
+
.card-tag-no {
|
| 274 |
+
background: rgba(100,116,139,0.15);
|
| 275 |
+
color: var(--muted);
|
| 276 |
+
}
|
| 277 |
+
|
| 278 |
+
/* ββ Figure images βββββββββββββββββββββββββ */
|
| 279 |
+
.figures-grid {
|
| 280 |
+
display: grid;
|
| 281 |
+
grid-template-columns: repeat(auto-fit, minmax(420px, 1fr));
|
| 282 |
+
gap: 20px;
|
| 283 |
+
}
|
| 284 |
+
.figure-card {
|
| 285 |
+
background: var(--surface);
|
| 286 |
+
border: 1px solid var(--border);
|
| 287 |
+
border-radius: var(--radius);
|
| 288 |
+
overflow: hidden;
|
| 289 |
+
}
|
| 290 |
+
.figure-card img {
|
| 291 |
+
width: 100%;
|
| 292 |
+
display: block;
|
| 293 |
+
background: #fff;
|
| 294 |
+
}
|
| 295 |
+
.figure-caption {
|
| 296 |
+
padding: 12px 16px;
|
| 297 |
+
font-size: 0.82rem;
|
| 298 |
+
color: var(--muted);
|
| 299 |
+
text-align: center;
|
| 300 |
+
}
|
| 301 |
+
|
| 302 |
+
/* ββ Dialect pills βββββββββββββββββββββββββ */
|
| 303 |
+
.dialect-list {
|
| 304 |
+
display: flex;
|
| 305 |
+
flex-wrap: wrap;
|
| 306 |
+
gap: 10px;
|
| 307 |
+
margin-top: 8px;
|
| 308 |
+
}
|
| 309 |
+
.dialect-pill {
|
| 310 |
+
background: var(--surface2);
|
| 311 |
+
border: 1px solid var(--border);
|
| 312 |
+
border-radius: 99px;
|
| 313 |
+
padding: 6px 14px;
|
| 314 |
+
font-size: 0.82rem;
|
| 315 |
+
display: flex;
|
| 316 |
+
align-items: center;
|
| 317 |
+
gap: 6px;
|
| 318 |
+
}
|
| 319 |
+
.dialect-pill .bar {
|
| 320 |
+
height: 4px;
|
| 321 |
+
border-radius: 99px;
|
| 322 |
+
background: var(--accent);
|
| 323 |
+
min-width: 20px;
|
| 324 |
+
}
|
| 325 |
+
|
| 326 |
+
/* ββ Scoring methodology βββββββββββββββββββ */
|
| 327 |
+
.method-steps {
|
| 328 |
+
display: grid;
|
| 329 |
+
grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
|
| 330 |
+
gap: 16px;
|
| 331 |
+
}
|
| 332 |
+
.step {
|
| 333 |
+
background: var(--surface);
|
| 334 |
+
border: 1px solid var(--border);
|
| 335 |
+
border-radius: var(--radius);
|
| 336 |
+
padding: 20px;
|
| 337 |
+
}
|
| 338 |
+
.step-num {
|
| 339 |
+
width: 32px;
|
| 340 |
+
height: 32px;
|
| 341 |
+
border-radius: 50%;
|
| 342 |
+
background: var(--accent);
|
| 343 |
+
color: #fff;
|
| 344 |
+
font-weight: 800;
|
| 345 |
+
font-size: 0.9rem;
|
| 346 |
+
display: flex;
|
| 347 |
+
align-items: center;
|
| 348 |
+
justify-content: center;
|
| 349 |
+
margin-bottom: 12px;
|
| 350 |
+
}
|
| 351 |
+
.step h4 {
|
| 352 |
+
font-size: 0.92rem;
|
| 353 |
+
font-weight: 700;
|
| 354 |
+
margin-bottom: 6px;
|
| 355 |
+
}
|
| 356 |
+
.step p {
|
| 357 |
+
font-size: 0.82rem;
|
| 358 |
+
color: var(--muted);
|
| 359 |
+
}
|
| 360 |
+
|
| 361 |
+
/* ββ Team ββββββββββββββββββββββββββββββββββ */
|
| 362 |
+
.team-grid {
|
| 363 |
+
display: grid;
|
| 364 |
+
grid-template-columns: repeat(auto-fit, minmax(220px, 1fr));
|
| 365 |
+
gap: 20px;
|
| 366 |
+
}
|
| 367 |
+
.team-card {
|
| 368 |
+
background: var(--surface);
|
| 369 |
+
border: 1px solid var(--border);
|
| 370 |
+
border-radius: var(--radius);
|
| 371 |
+
padding: 24px;
|
| 372 |
+
text-align: center;
|
| 373 |
+
transition: border-color 0.2s;
|
| 374 |
+
}
|
| 375 |
+
.team-card:hover { border-color: var(--accent); }
|
| 376 |
+
.team-avatar {
|
| 377 |
+
width: 56px;
|
| 378 |
+
height: 56px;
|
| 379 |
+
border-radius: 50%;
|
| 380 |
+
background: linear-gradient(135deg, var(--accent), var(--accent2));
|
| 381 |
+
display: flex;
|
| 382 |
+
align-items: center;
|
| 383 |
+
justify-content: center;
|
| 384 |
+
font-size: 1.4rem;
|
| 385 |
+
font-weight: 800;
|
| 386 |
+
color: #fff;
|
| 387 |
+
margin: 0 auto 14px;
|
| 388 |
+
}
|
| 389 |
+
.team-name { font-weight: 700; font-size: 1rem; margin-bottom: 4px; }
|
| 390 |
+
.team-role { font-size: 0.82rem; color: var(--muted); margin-bottom: 8px; }
|
| 391 |
+
.team-hf {
|
| 392 |
+
display: inline-flex;
|
| 393 |
+
align-items: center;
|
| 394 |
+
gap: 5px;
|
| 395 |
+
font-size: 0.78rem;
|
| 396 |
+
color: var(--accent);
|
| 397 |
+
text-decoration: none;
|
| 398 |
+
}
|
| 399 |
+
.team-hf:hover { text-decoration: underline; }
|
| 400 |
+
|
| 401 |
+
/* ββ Dataset callout βββββββββββββββββββββββ */
|
| 402 |
+
.dataset-callout {
|
| 403 |
+
background: linear-gradient(135deg, rgba(16,185,129,0.08), rgba(99,102,241,0.08));
|
| 404 |
+
border: 1px solid rgba(16,185,129,0.3);
|
| 405 |
+
border-radius: var(--radius);
|
| 406 |
+
padding: 28px 32px;
|
| 407 |
+
display: flex;
|
| 408 |
+
align-items: center;
|
| 409 |
+
gap: 24px;
|
| 410 |
+
flex-wrap: wrap;
|
| 411 |
+
}
|
| 412 |
+
.dataset-callout .ds-icon { font-size: 2.5rem; }
|
| 413 |
+
.dataset-callout h3 { font-size: 1.1rem; font-weight: 700; margin-bottom: 4px; }
|
| 414 |
+
.dataset-callout p { font-size: 0.88rem; color: var(--muted); }
|
| 415 |
+
|
| 416 |
+
/* ββ Score note ββββββββββββββββββββββββββββ */
|
| 417 |
+
.note-box {
|
| 418 |
+
background: rgba(99,102,241,0.08);
|
| 419 |
+
border: 1px solid rgba(99,102,241,0.25);
|
| 420 |
+
border-radius: 8px;
|
| 421 |
+
padding: 14px 18px;
|
| 422 |
+
font-size: 0.85rem;
|
| 423 |
+
color: var(--muted);
|
| 424 |
+
margin-top: 20px;
|
| 425 |
+
}
|
| 426 |
+
.note-box strong { color: var(--text); }
|
| 427 |
+
|
| 428 |
+
/* ββ Footer ββββββββββββββββββββββββββββββββ */
|
| 429 |
+
footer {
|
| 430 |
+
border-top: 1px solid var(--border);
|
| 431 |
+
padding: 32px 24px;
|
| 432 |
+
text-align: center;
|
| 433 |
+
font-size: 0.82rem;
|
| 434 |
+
color: var(--muted);
|
| 435 |
+
}
|
| 436 |
+
footer a { color: var(--accent); text-decoration: none; }
|
| 437 |
+
footer a:hover { text-decoration: underline; }
|
| 438 |
+
|
| 439 |
+
@media (max-width: 600px) {
|
| 440 |
+
.hero { padding: 40px 16px 32px; }
|
| 441 |
+
.figures-grid { grid-template-columns: 1fr; }
|
| 442 |
+
.dataset-callout { flex-direction: column; gap: 12px; }
|
| 443 |
+
}
|
| 444 |
+
</style>
|
| 445 |
+
</head>
|
| 446 |
+
<body>
|
| 447 |
+
|
| 448 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 449 |
+
HERO
|
| 450 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 451 |
+
<div class="hero">
|
| 452 |
+
<div class="hero-logo">
|
| 453 |
+
<img src="haakkim-logo-withname.png" alt="Haakkim" onerror="this.style.display='none'" />
|
| 454 |
+
</div>
|
| 455 |
+
<div class="hero-ar">ΨΩΩΩΩΩ
</div>
|
| 456 |
+
<p>An open arena-style human preference evaluation platform for Arabic large language models β built from the ground up for Arabic.</p>
|
| 457 |
+
<div class="hero-links">
|
| 458 |
+
<a href="https://haakkim.tech" class="btn btn-primary">
|
| 459 |
+
π Live Platform
|
| 460 |
+
</a>
|
| 461 |
+
<a href="https://haakkim.tech/#leaderboard" class="btn btn-outline">
|
| 462 |
+
π Leaderboard
|
| 463 |
+
</a>
|
| 464 |
+
<a href="https://huggingface.co/datasets/Haakkim/Haakkim-1.0v" class="btn btn-dataset">
|
| 465 |
+
π¦ Dataset v1.0
|
| 466 |
+
</a>
|
| 467 |
+
</div>
|
| 468 |
+
</div>
|
| 469 |
+
|
| 470 |
+
<div class="container">
|
| 471 |
+
|
| 472 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 473 |
+
SNAPSHOT STATS
|
| 474 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 475 |
+
<section>
|
| 476 |
+
<div class="section-title">Current Snapshot β v1.0</div>
|
| 477 |
+
<div class="section-sub">Statistics from the first public release of Haakkim battle data</div>
|
| 478 |
+
<div class="stats-grid">
|
| 479 |
+
<div class="stat-card">
|
| 480 |
+
<div class="stat-value">1,273</div>
|
| 481 |
+
<div class="stat-label">Total Battles</div>
|
| 482 |
+
</div>
|
| 483 |
+
<div class="stat-card">
|
| 484 |
+
<div class="stat-value">831</div>
|
| 485 |
+
<div class="stat-label">BT-Ranked Battles</div>
|
| 486 |
+
</div>
|
| 487 |
+
<div class="stat-card">
|
| 488 |
+
<div class="stat-value">67</div>
|
| 489 |
+
<div class="stat-label">Models Ranked</div>
|
| 490 |
+
</div>
|
| 491 |
+
<div class="stat-card">
|
| 492 |
+
<div class="stat-value">11</div>
|
| 493 |
+
<div class="stat-label">Arabic Dialects</div>
|
| 494 |
+
</div>
|
| 495 |
+
<div class="stat-card">
|
| 496 |
+
<div class="stat-value">465</div>
|
| 497 |
+
<div class="stat-label">ESS (Clamped)</div>
|
| 498 |
+
</div>
|
| 499 |
+
<div class="stat-card">
|
| 500 |
+
<div class="stat-value">0.35</div>
|
| 501 |
+
<div class="stat-label">Graph Density</div>
|
| 502 |
+
</div>
|
| 503 |
+
</div>
|
| 504 |
+
</section>
|
| 505 |
+
|
| 506 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 507 |
+
MSA LEADERBOARD TOP 10
|
| 508 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 509 |
+
<section>
|
| 510 |
+
<div class="section-title">MSA Leaderboard β Top 10</div>
|
| 511 |
+
<div class="section-sub">BradleyβTerry scores (1000-centered log-odds). Full 67-model leaderboard at <a href="https://haakkim.tech/#leaderboard" style="color:var(--accent)">haakkim.tech</a></div>
|
| 512 |
+
<div class="table-wrap">
|
| 513 |
+
<table>
|
| 514 |
+
<thead>
|
| 515 |
+
<tr>
|
| 516 |
+
<th>Rank</th>
|
| 517 |
+
<th>Model</th>
|
| 518 |
+
<th>BT Score</th>
|
| 519 |
+
<th>95% CI</th>
|
| 520 |
+
<th>Battles</th>
|
| 521 |
+
</tr>
|
| 522 |
+
</thead>
|
| 523 |
+
<tbody>
|
| 524 |
+
<tr>
|
| 525 |
+
<td><span class="rank-badge rank-1">1</span></td>
|
| 526 |
+
<td><span class="model-org">mistralai/</span><span class="model-name">ministral-3b-2512</span></td>
|
| 527 |
+
<td><span class="score-val">1001.75</span></td>
|
| 528 |
+
<td><span class="ci-val">[1001.20, 1002.93]</span></td>
|
| 529 |
+
<td><span class="battles-badge">40</span></td>
|
| 530 |
+
</tr>
|
| 531 |
+
<tr>
|
| 532 |
+
<td><span class="rank-badge rank-2">2</span></td>
|
| 533 |
+
<td><span class="model-org">mistralai/</span><span class="model-name">ministral-8b-2512</span></td>
|
| 534 |
+
<td><span class="score-val">1001.61</span></td>
|
| 535 |
+
<td><span class="ci-val">[1000.72, 1002.97]</span></td>
|
| 536 |
+
<td><span class="battles-badge">43</span></td>
|
| 537 |
+
</tr>
|
| 538 |
+
<tr>
|
| 539 |
+
<td><span class="rank-badge rank-3">3</span></td>
|
| 540 |
+
<td><span class="model-org">Qwen/</span><span class="model-name">Qwen3-235B-A22B-Thinking-2507</span></td>
|
| 541 |
+
<td><span class="score-val">1001.21</span></td>
|
| 542 |
+
<td><span class="ci-val">[1000.47, 1002.00]</span></td>
|
| 543 |
+
<td><span class="battles-badge">38</span></td>
|
| 544 |
+
</tr>
|
| 545 |
+
<tr>
|
| 546 |
+
<td><span class="rank-badge rank-other">4</span></td>
|
| 547 |
+
<td><span class="model-org">Qwen/</span><span class="model-name">Qwen3-30B-A3B-Instruct-2507</span></td>
|
| 548 |
+
<td><span class="score-val">1001.14</span></td>
|
| 549 |
+
<td><span class="ci-val">[999.96, 1002.83]</span></td>
|
| 550 |
+
<td><span class="battles-badge">31</span></td>
|
| 551 |
+
</tr>
|
| 552 |
+
<tr>
|
| 553 |
+
<td><span class="rank-badge rank-other">5</span></td>
|
| 554 |
+
<td><span class="model-org">deepseek/</span><span class="model-name">deepseek-v3.2-exp</span></td>
|
| 555 |
+
<td><span class="score-val">1001.13</span></td>
|
| 556 |
+
<td><span class="ci-val">[1000.27, 1002.16]</span></td>
|
| 557 |
+
<td><span class="battles-badge">38</span></td>
|
| 558 |
+
</tr>
|
| 559 |
+
<tr>
|
| 560 |
+
<td><span class="rank-badge rank-other">6</span></td>
|
| 561 |
+
<td><span class="model-org">deepseek/</span><span class="model-name">deepseek-v3.1</span></td>
|
| 562 |
+
<td><span class="score-val">1000.99</span></td>
|
| 563 |
+
<td><span class="ci-val">[999.81, 1002.07]</span></td>
|
| 564 |
+
<td><span class="battles-badge">29</span></td>
|
| 565 |
+
</tr>
|
| 566 |
+
<tr>
|
| 567 |
+
<td><span class="rank-badge rank-other">7</span></td>
|
| 568 |
+
<td><span class="model-org">Qwen/</span><span class="model-name">Qwen3-235B-A22B-Instruct-2507</span></td>
|
| 569 |
+
<td><span class="score-val">1000.98</span></td>
|
| 570 |
+
<td><span class="ci-val">[1000.12, 1002.08]</span></td>
|
| 571 |
+
<td><span class="battles-badge">39</span></td>
|
| 572 |
+
</tr>
|
| 573 |
+
<tr>
|
| 574 |
+
<td><span class="rank-badge rank-other">8</span></td>
|
| 575 |
+
<td><span class="model-org">deepseek/</span><span class="model-name">deepseek-r1-0528</span></td>
|
| 576 |
+
<td><span class="score-val">1000.93</span></td>
|
| 577 |
+
<td><span class="ci-val">[1000.10, 1002.14]</span></td>
|
| 578 |
+
<td><span class="battles-badge">38</span></td>
|
| 579 |
+
</tr>
|
| 580 |
+
<tr>
|
| 581 |
+
<td><span class="rank-badge rank-other">9</span></td>
|
| 582 |
+
<td><span class="model-org">openai/</span><span class="model-name">gpt-oss-120b</span></td>
|
| 583 |
+
<td><span class="score-val">1000.93</span></td>
|
| 584 |
+
<td><span class="ci-val">[1000.04, 1002.58]</span></td>
|
| 585 |
+
<td><span class="battles-badge">25</span></td>
|
| 586 |
+
</tr>
|
| 587 |
+
<tr>
|
| 588 |
+
<td><span class="rank-badge rank-other">10</span></td>
|
| 589 |
+
<td><span class="model-org">deepseek/</span><span class="model-name">deepseek-v3.2</span></td>
|
| 590 |
+
<td><span class="score-val">1000.89</span></td>
|
| 591 |
+
<td><span class="ci-val">[999.86, 1002.25]</span></td>
|
| 592 |
+
<td><span class="battles-badge">31</span></td>
|
| 593 |
+
</tr>
|
| 594 |
+
</tbody>
|
| 595 |
+
</table>
|
| 596 |
+
</div>
|
| 597 |
+
<div class="note-box">
|
| 598 |
+
<strong>Score scale:</strong> Haakkim uses unscaled log-odds units centered at 1000 β a 1-point gap corresponds to win odds of eΒΉ β 2.7:1. This produces a ~4-point spread across 67 models. Chatbot Arena-style Elo scaling (Γ173.7) would show the same ranking with hundreds-of-points spreads; both conventions encode identical win probabilities.
|
| 599 |
+
</div>
|
| 600 |
+
</section>
|
| 601 |
+
|
| 602 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 603 |
+
FIGURES
|
| 604 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 605 |
+
<section>
|
| 606 |
+
<div class="section-title">Analysis Figures</div>
|
| 607 |
+
<div class="section-sub">Regenerated from the live database β v1.0 snapshot</div>
|
| 608 |
+
<div class="figures-grid">
|
| 609 |
+
<div class="figure-card">
|
| 610 |
+
<img src="bt_leaderboard_ci_top20.png" alt="BT Leaderboard Top 20 with Confidence Intervals" />
|
| 611 |
+
<div class="figure-caption">BT Leaderboard β Top 20 models with 95% bootstrap confidence intervals</div>
|
| 612 |
+
</div>
|
| 613 |
+
<div class="figure-card">
|
| 614 |
+
<img src="dialect_distribution.png" alt="Dialect Distribution" />
|
| 615 |
+
<div class="figure-caption">Battle distribution across 11 Arabic dialect varieties</div>
|
| 616 |
+
</div>
|
| 617 |
+
<div class="figure-card">
|
| 618 |
+
<img src="vote_outcome_distribution_ranked.png" alt="Vote Outcome Distribution" />
|
| 619 |
+
<div class="figure-caption">Vote outcome distribution for ranked arena battles</div>
|
| 620 |
+
</div>
|
| 621 |
+
<div class="figure-card">
|
| 622 |
+
<img src="haakkim_overview.png" alt="Platform Overview" />
|
| 623 |
+
<div class="figure-caption">Haakkim platform overview and evaluation pipeline</div>
|
| 624 |
+
</div>
|
| 625 |
+
</div>
|
| 626 |
+
</section>
|
| 627 |
+
|
| 628 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 629 |
+
DIALECT COVERAGE
|
| 630 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 631 |
+
<section>
|
| 632 |
+
<div class="section-title">Arabic Dialect Coverage</div>
|
| 633 |
+
<div class="section-sub">11 varieties β from Modern Standard Arabic to regional dialects across the Arab world</div>
|
| 634 |
+
<div class="dialect-list">
|
| 635 |
+
<div class="dialect-pill">MSA <div class="bar" style="width:80px;background:#6366f1"></div> 77.5%</div>
|
| 636 |
+
<div class="dialect-pill">Tunisian <div class="bar" style="width:22px;background:#8b5cf6"></div> 9.0%</div>
|
| 637 |
+
<div class="dialect-pill">Saudi <div class="bar" style="width:16px;background:#a78bfa"></div> 6.5%</div>
|
| 638 |
+
<div class="dialect-pill">Egyptian <div class="bar" style="width:9px;background:#c4b5fd"></div> 3.5%</div>
|
| 639 |
+
<div class="dialect-pill">Levantine <div class="bar" style="width:6px"></div> 1.7%</div>
|
| 640 |
+
<div class="dialect-pill">Sudanese <div class="bar" style="width:4px"></div> 0.9%</div>
|
| 641 |
+
<div class="dialect-pill">Omani <div class="bar" style="width:3px"></div> 0.4%</div>
|
| 642 |
+
<div class="dialect-pill">Iraqi <div class="bar" style="width:2px"></div> 0.2%</div>
|
| 643 |
+
<div class="dialect-pill">Moroccan <div class="bar" style="width:2px"></div> <0.1%</div>
|
| 644 |
+
<div class="dialect-pill">Libyan <div class="bar" style="width:2px"></div> <0.1%</div>
|
| 645 |
+
<div class="dialect-pill">Algerian <div class="bar" style="width:2px"></div> <0.1%</div>
|
| 646 |
+
</div>
|
| 647 |
+
</section>
|
| 648 |
+
|
| 649 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 650 |
+
EVALUATION MODES
|
| 651 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 652 |
+
<section>
|
| 653 |
+
<div class="section-title">Evaluation Modes</div>
|
| 654 |
+
<div class="section-sub">Three ways to compare Arabic LLMs β only Ranked Arena feeds the official leaderboard</div>
|
| 655 |
+
<div class="cards-grid">
|
| 656 |
+
<div class="card">
|
| 657 |
+
<div class="card-icon">βοΈ</div>
|
| 658 |
+
<h3>Ranked Arena</h3>
|
| 659 |
+
<p>Random model pairing, single-turn MSA, matched system instruction. Results feed the official BradleyβTerry leaderboard.</p>
|
| 660 |
+
<span class="card-tag">β BT Leaderboard</span>
|
| 661 |
+
</div>
|
| 662 |
+
<div class="card">
|
| 663 |
+
<div class="card-icon">βοΈ</div>
|
| 664 |
+
<h3>Side-by-Side</h3>
|
| 665 |
+
<p>User-selected model pair, any dialect. Useful for targeted comparisons but excluded from ranked scoring to prevent selection bias.</p>
|
| 666 |
+
<span class="card-tag card-tag-no">Win-rate only</span>
|
| 667 |
+
</div>
|
| 668 |
+
<div class="card">
|
| 669 |
+
<div class="card-icon">β</div>
|
| 670 |
+
<h3>10 Questions</h3>
|
| 671 |
+
<p>Fixed Arabic prompt pool, any dialect. Provides consistent benchmarking within a curated set of questions.</p>
|
| 672 |
+
<span class="card-tag card-tag-no">Win-rate only</span>
|
| 673 |
+
</div>
|
| 674 |
+
</div>
|
| 675 |
+
</section>
|
| 676 |
+
|
| 677 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 678 |
+
SCORING METHODOLOGY
|
| 679 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 680 |
+
<section>
|
| 681 |
+
<div class="section-title">Scoring Methodology</div>
|
| 682 |
+
<div class="section-sub">Statistically rigorous BradleyβTerry model with four key components</div>
|
| 683 |
+
<div class="method-steps">
|
| 684 |
+
<div class="step">
|
| 685 |
+
<div class="step-num">1</div>
|
| 686 |
+
<h4>Inverse-Probability Weighting</h4>
|
| 687 |
+
<p>Corrects for non-uniform model exposure using Ξ΅-greedy adaptive sampling weights, clamped to [P1, P99].</p>
|
| 688 |
+
</div>
|
| 689 |
+
<div class="step">
|
| 690 |
+
<div class="step-num">2</div>
|
| 691 |
+
<h4>Bootstrap Confidence Intervals</h4>
|
| 692 |
+
<p>200 vote-level resamples per run to produce 95% CIs on every model's BT score.</p>
|
| 693 |
+
</div>
|
| 694 |
+
<div class="step">
|
| 695 |
+
<div class="step-num">3</div>
|
| 696 |
+
<h4>Rankability Gate</h4>
|
| 697 |
+
<p>BT scores published only when the comparison graph is fully connected and ESS is sufficient; otherwise win-rate fallback is shown.</p>
|
| 698 |
+
</div>
|
| 699 |
+
<div class="step">
|
| 700 |
+
<div class="step-num">4</div>
|
| 701 |
+
<h4>Log-odds Scale</h4>
|
| 702 |
+
<p>1000-centered unscaled log-odds. A 1-point gap β 2.7:1 win odds. Full reproducibility: pipeline and dataset are open.</p>
|
| 703 |
+
</div>
|
| 704 |
+
</div>
|
| 705 |
+
</section>
|
| 706 |
+
|
| 707 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 708 |
+
DATASET CALLOUT
|
| 709 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 710 |
+
<section>
|
| 711 |
+
<div class="dataset-callout">
|
| 712 |
+
<div class="ds-icon">π¦</div>
|
| 713 |
+
<div style="flex:1">
|
| 714 |
+
<h3>Haakkim/Haakkim-1.0v β Battle Dataset</h3>
|
| 715 |
+
<p>1,273 battle records (Parquet, PII-scrubbed). Includes voted comparisons and skipped battles across all 11 dialects and 3 evaluation modes. Full conversation transcripts, sampling weights, category annotations.</p>
|
| 716 |
+
</div>
|
| 717 |
+
<a href="https://huggingface.co/datasets/Haakkim/Haakkim-1.0v" class="btn btn-dataset">View Dataset β</a>
|
| 718 |
+
</div>
|
| 719 |
+
</section>
|
| 720 |
+
|
| 721 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 722 |
+
TEAM
|
| 723 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 724 |
+
<section>
|
| 725 |
+
<div class="section-title">Team</div>
|
| 726 |
+
<div class="section-sub">College of Computing, Umm Al-Qura University β Mecca, Saudi Arabia</div>
|
| 727 |
+
<div class="team-grid">
|
| 728 |
+
<div class="team-card">
|
| 729 |
+
<div class="team-avatar">MM</div>
|
| 730 |
+
<div class="team-name">Mourad Mars</div>
|
| 731 |
+
<div class="team-role">Principal Investigator</div>
|
| 732 |
+
<a href="https://huggingface.co/mouradmars" class="team-hf">π€ mouradmars</a>
|
| 733 |
+
</div>
|
| 734 |
+
<div class="team-card">
|
| 735 |
+
<div class="team-avatar">HB</div>
|
| 736 |
+
<div class="team-name">Hassan Barmandah</div>
|
| 737 |
+
<div class="team-role">AI Researcher</div>
|
| 738 |
+
<a href="https://huggingface.co/HassanB4" class="team-hf">π€ HassanB4</a>
|
| 739 |
+
</div>
|
| 740 |
+
<div class="team-card">
|
| 741 |
+
<div class="team-avatar">AA</div>
|
| 742 |
+
<div class="team-name">Abdulrhman Alassaf</div>
|
| 743 |
+
<div class="team-role">Software Engineer</div>
|
| 744 |
+
<span style="font-size:0.78rem;color:var(--muted)">Umm Al-Qura University</span>
|
| 745 |
+
</div>
|
| 746 |
+
</div>
|
| 747 |
+
</section>
|
| 748 |
+
|
| 749 |
+
<!-- βββββββββββββββββββββββββββββββββββββββββββββββββββ
|
| 750 |
+
CITATION
|
| 751 |
+
ββββββββββββββββββββββββββββββββββββββββββββββββββββ -->
|
| 752 |
+
<section>
|
| 753 |
+
<div class="section-title">Citation</div>
|
| 754 |
+
<div class="section-sub">If you use Haakkim or this dataset in your research, please cite:</div>
|
| 755 |
+
<pre style="background:var(--surface);border:1px solid var(--border);border-radius:var(--radius);padding:20px 24px;font-size:0.82rem;line-height:1.7;overflow-x:auto;color:var(--muted)"><code>@misc{mars2026haakkim,
|
| 756 |
+
title = {Haakkim: An Arena-Style Human Preference Evaluation Platform for Arabic {LLMs}},
|
| 757 |
+
author = {Mars, Mourad and Barmandah, Hassan and Alassaf, Abdulrhman},
|
| 758 |
+
year = {2026},
|
| 759 |
+
howpublished = {\url{https://huggingface.co/datasets/Haakkim/Haakkim-1.0v}},
|
| 760 |
+
note = {College of Computing, Umm Al-Qura University, Mecca, Saudi Arabia}
|
| 761 |
+
}</code></pre>
|
| 762 |
+
</section>
|
| 763 |
+
|
| 764 |
+
</div><!-- /container -->
|
| 765 |
+
|
| 766 |
+
<footer>
|
| 767 |
+
<div>
|
| 768 |
+
<a href="https://haakkim.tech">haakkim.tech</a> Β·
|
| 769 |
+
<a href="https://haakkim.tech/#leaderboard">Leaderboard</a> Β·
|
| 770 |
+
<a href="https://huggingface.co/datasets/Haakkim/Haakkim-1.0v">Dataset</a>
|
| 771 |
+
</div>
|
| 772 |
+
<div style="margin-top:8px">College of Computing, Umm Al-Qura University Β· Mecca, Saudi Arabia Β· CC BY 4.0</div>
|
| 773 |
+
</footer>
|
| 774 |
+
|
| 775 |
+
</body>
|
| 776 |
+
</html>
|