KyleHessling1's picture
Initial upload: Jackrong DeepSeek-V4-Flash 9B eval with base 9B comparison
bcab855 verified
Raw
History Blame Contribute Delete
11.7 kB
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<title>Qwen3.5-9B-DeepSeek-V4-Flash — Evaluation Suite</title>
<meta name="viewport" content="width=device-width,initial-scale=1">
<style>
:root { color-scheme: dark; }
* { box-sizing: border-box; }
body {
font-family: -apple-system, BlinkMacSystemFont, "SF Pro Text", "Inter", system-ui, sans-serif;
background: radial-gradient(ellipse at top, #0a1f1a 0%, #050a09 60%);
color: #e6f3ee;
padding: 4rem 1.5rem 6rem;
margin: 0;
min-height: 100vh;
line-height: 1.55;
}
.wrap { max-width: 920px; margin: 0 auto; }
header { margin-bottom: 3rem; }
.eyebrow {
color: #6fd4a8;
font-size: 0.78rem;
letter-spacing: 0.22em;
text-transform: uppercase;
font-weight: 600;
}
h1 {
font-size: clamp(1.8rem, 4vw, 2.6rem);
font-weight: 700;
letter-spacing: -0.025em;
margin: 0.35rem 0 0.6rem;
background: linear-gradient(135deg, #ffffff 0%, #6fd4a8 100%);
-webkit-background-clip: text;
background-clip: text;
-webkit-text-fill-color: transparent;
}
.sub { color: #95b3a8; max-width: 60ch; }
.stats {
display: grid;
grid-template-columns: repeat(auto-fit, minmax(150px, 1fr));
gap: 1rem;
margin: 2rem 0 3rem;
padding: 1.25rem;
background: rgba(255,255,255,0.03);
border: 1px solid rgba(255,255,255,0.06);
border-radius: 14px;
}
.stat b {
display: block;
font-size: 1.6rem;
font-weight: 700;
letter-spacing: -0.02em;
color: #ffffff;
}
.stat span {
color: #7e9d92;
font-size: 0.78rem;
text-transform: uppercase;
letter-spacing: 0.12em;
}
h2 {
font-size: 0.85rem;
color: #93c8b1;
text-transform: uppercase;
letter-spacing: 0.14em;
font-weight: 600;
margin: 3rem 0 1rem;
padding-bottom: 0.6rem;
border-bottom: 1px solid rgba(255,255,255,0.07);
}
.grid {
display: grid;
grid-template-columns: repeat(auto-fill, minmax(280px, 1fr));
gap: 0.9rem;
}
.card {
display: block;
padding: 1.1rem 1.2rem;
background: rgba(255,255,255,0.035);
border: 1px solid rgba(255,255,255,0.07);
border-radius: 11px;
color: inherit;
text-decoration: none;
transition: transform 0.22s ease, border-color 0.22s ease, background 0.22s ease;
}
.card:hover {
transform: translateY(-2px);
border-color: rgba(111, 212, 168, 0.45);
background: rgba(111, 212, 168, 0.07);
}
.card strong {
display: block;
font-weight: 600;
font-size: 0.98rem;
color: #f1faf5;
margin-bottom: 0.25rem;
}
.card span {
color: #88a89c;
font-size: 0.82rem;
}
.meta {
font-family: ui-monospace, SFMono-Regular, Menlo, Consolas, monospace;
font-size: 0.72rem;
color: #6e8c81;
margin-top: 0.6rem;
}
.card-ab {
padding: 1.1rem 1.2rem;
background: rgba(255,255,255,0.035);
border: 1px solid rgba(255,255,255,0.07);
border-radius: 11px;
}
.card-head strong {
display: block;
font-weight: 600;
font-size: 0.98rem;
color: #f1faf5;
margin-bottom: 0.25rem;
}
.card-head span {
color: #88a89c;
font-size: 0.82rem;
}
.ab-row {
display: grid;
grid-template-columns: 1fr 1fr;
gap: 0.5rem;
margin: 0.85rem 0 0.55rem;
}
.ab-link {
display: flex; flex-direction: column; align-items: flex-start;
padding: 0.55rem 0.7rem;
border-radius: 7px;
text-decoration: none;
transition: transform 0.18s ease, background 0.18s ease, border-color 0.18s ease;
border: 1px solid transparent;
}
.ab-link:hover { transform: translateY(-1px); }
.ab-link.distill {
background: linear-gradient(135deg, rgba(45, 163, 119, 0.22) 0%, rgba(111, 212, 168, 0.18) 100%);
border-color: rgba(111, 212, 168, 0.35);
}
.ab-link.distill:hover {
background: linear-gradient(135deg, rgba(45, 163, 119, 0.34) 0%, rgba(111, 212, 168, 0.30) 100%);
border-color: rgba(111, 212, 168, 0.6);
}
.ab-link.base {
background: rgba(255,255,255,0.04);
border-color: rgba(255,255,255,0.08);
}
.ab-link.base:hover {
background: rgba(255,255,255,0.07);
border-color: rgba(255,255,255,0.18);
}
.ab-label {
font-size: 0.7rem; letter-spacing: 0.14em;
text-transform: uppercase; font-weight: 700;
}
.ab-link.distill .ab-label { color: #9bedc9; }
.ab-link.base .ab-label { color: #c4d8d0; }
.ab-meta {
font-family: ui-monospace, SFMono-Regular, Menlo, Consolas, monospace;
font-size: 0.72rem;
color: #b3e6cf;
margin-top: 0.18rem;
}
.ab-link.base .ab-meta { color: #9bb5ab; }
.cta-row { display: flex; flex-wrap: wrap; gap: 0.7rem; align-items: center; margin-top: 1.5rem; }
.follow-link, .model-link {
display: inline-flex; align-items: center; gap: 0.55rem;
padding: 0.85rem 1.2rem;
background: rgba(255,255,255,0.055);
border: 1px solid rgba(255,255,255,0.1);
color: #e6f3ee; border-radius: 10px;
text-decoration: none; font-weight: 600; font-size: 0.95rem;
transition: transform 0.2s ease, background 0.2s ease, border-color 0.2s ease;
}
.follow-link:hover, .model-link:hover {
transform: translateY(-1px);
background: rgba(255,255,255,0.09);
border-color: rgba(255,255,255,0.2);
}
.model-link {
background: linear-gradient(135deg, #2da377 0%, #6fd4a8 100%);
border: none;
}
.byline {
color: #9bc7b3;
font-size: 0.92rem;
margin: 0.1rem 0 0.9rem;
letter-spacing: 0.01em;
}
.byline a { color: inherit; text-decoration: none;
border-bottom: 1px solid rgba(155, 199, 179, 0.25); }
.byline a:hover { border-bottom-color: #9bc7b3; }
footer {
margin-top: 5rem;
color: #557067;
font-size: 0.82rem;
text-align: center;
}
</style>
</head>
<body>
<div class="wrap">
<header>
<div class="eyebrow">Model Evaluation</div>
<h1>Qwen3.5-9B-DeepSeek-V4-Flash · Q5_K_M</h1>
<p class="byline">eval by <a href="https://x.com/KyleHessling1" target="_blank" rel="noopener">Kyle Hessling</a> · model by <a href="https://huggingface.co/Jackrong" target="_blank" rel="noopener">Jackrong</a></p>
<p class="sub">A 9B Qwen 3.5 finetune distilled from DeepSeek V4, benchmarked side-by-side with the official <a href="https://huggingface.co/Qwen/Qwen3.5-9B" target="_blank" rel="noopener" style="color:#9bedc9">Qwen3.5-9B</a> base at the same Q5_K_M quant. <strong>The distill thinks tighter on agentic tasks AND produces visibly better one-shot creative front-end design — open the A/B buttons on the cards below and judge for yourself.</strong> Tool-calling and throughput tie. Full breakdown in the report.</p>
<div class="cta-row">
<a href="report.html" class="model-link">Read the full report →</a>
<a href="https://huggingface.co/Jackrong/Qwen3.5-9B-DeepSeek-V4-Flash" target="_blank" rel="noopener" class="follow-link">Model on Hugging Face</a>
<a href="https://huggingface.co/Jackrong" target="_blank" rel="noopener" class="follow-link">More from Jackrong →</a>
<a href="https://x.com/KyleHessling1" target="_blank" rel="noopener" class="follow-link">
<svg viewBox="0 0 24 24" width="15" height="15" fill="currentColor" aria-hidden="true"><path d="M18.244 2.25h3.308l-7.227 8.26 8.502 11.24H16.17l-5.214-6.817L4.99 21.75H1.68l7.73-8.835L1.254 2.25H8.08l4.713 6.231zm-1.161 17.52h1.833L7.084 4.126H5.117z"/></svg>
Follow @KyleHessling1
</a>
</div>
</header>
<div class="stats">
<div class="stat"><b>0 / 5</b><span>distill cap hits</span></div>
<div class="stat"><b>3 / 5</b><span>base cap hits</span></div>
<div class="stat"><b>2.2×</b><span>faster agentic</span></div>
<div class="stat"><b>5/6</b><span>tools (both)</span></div>
<div class="stat"><b>142</b><span>tok/s (both)</span></div>
</div>
<h2>Web design · open to preview</h2>
<div class="grid">
<div class="card-ab"><div class="card-head"><strong>SaaS landing page</strong><span>Prism — AI observability</span></div><div class="ab-row"><a class="ab-link distill" href="design_saas_landing.html"><span class="ab-label">Distill</span><span class="ab-meta">15347 tok · 109 s</span></a><a class="ab-link base" href="base_design_saas_landing.html"><span class="ab-label">Base</span><span class="ab-meta">9849 tok · 68 s</span></a></div><div class="meta">distill: 44.2 KB · same prompt, both runs</div></div>
<div class="card-ab"><div class="card-head"><strong>Analytics dashboard</strong><span>Light theme, emerald accent</span></div><div class="ab-row"><a class="ab-link distill" href="design_analytics_dashboard.html"><span class="ab-label">Distill</span><span class="ab-meta">13032 tok · 93 s</span></a><a class="ab-link base" href="base_design_analytics_dashboard.html"><span class="ab-label">Base</span><span class="ab-meta">13187 tok · 91 s</span></a></div><div class="meta">distill: 41.1 KB · same prompt, both runs</div></div>
<div class="card-ab"><div class="card-head"><strong>Designer portfolio</strong><span>Maya Chen — kinetic typography</span></div><div class="ab-row"><a class="ab-link distill" href="design_designer_portfolio.html"><span class="ab-label">Distill</span><span class="ab-meta">6213 tok · 44 s</span></a><a class="ab-link base" href="base_design_designer_portfolio.html"><span class="ab-label">Base</span><span class="ab-meta">5930 tok · 41 s</span></a></div><div class="meta">distill: 18.0 KB · same prompt, both runs</div></div>
<div class="card-ab"><div class="card-head"><strong>Pricing page</strong><span>3 tiers + animated toggle + FAQ</span></div><div class="ab-row"><a class="ab-link distill" href="design_pricing_page.html"><span class="ab-label">Distill</span><span class="ab-meta">8367 tok · 59 s</span></a><a class="ab-link base" href="base_design_pricing_page.html"><span class="ab-label">Base</span><span class="ab-meta">9503 tok · 65 s</span></a></div><div class="meta">distill: 25.6 KB · same prompt, both runs</div></div>
<div class="card-ab"><div class="card-head"><strong>Mobile app marketing</strong><span>Stillwater — CSS-only iPhone mock</span></div><div class="ab-row"><a class="ab-link distill" href="design_mobile_app_marketing.html"><span class="ab-label">Distill</span><span class="ab-meta">10161 tok · 72 s</span></a><a class="ab-link base" href="base_design_mobile_app_marketing.html"><span class="ab-label">Base</span><span class="ab-meta">32000 tok · 228 s ⚠</span></a></div><div class="meta">distill: 30.5 KB · same prompt, both runs</div></div>
</div>
<h2>Agentic reasoning · text output</h2>
<div class="grid">
<a class="card" href="agentic_code_debug.txt"><strong>Code debug (4 bugs)</strong><span>k-th smallest element</span><div class="meta">3170 tok · 22 s</div></a>
<a class="card" href="agentic_multi_step_planning.txt"><strong>Multi-step planning</strong><span>URL shortener deploy plan</span><div class="meta">2899 tok · 20 s</div></a>
<a class="card" href="agentic_self_critique.txt"><strong>Self-critique loop</strong><span>Palindrome · O(n³) → O(n²)</span><div class="meta">1969 tok · 14 s</div></a>
<a class="card" href="agentic_structured_extraction.txt"><strong>Structured JSON extraction</strong><span>Calendar + roster from prose</span><div class="meta">4353 tok · 30 s</div></a>
<a class="card" href="agentic_tool_use_json.txt"><strong>Tool-use planning</strong><span>Weather + flights + hotel</span><div class="meta">1415 tok · 10 s</div></a>
</div>
<footer>
Generated from Qwen3.5-9B-DeepSeek-V4-Flash · Q5_K_M via llama.cpp · self-hosted on a single RTX 5090
</footer>
</div>
</body>
</html>