pulpie / index.html
not-lain's picture
use gradio server (#1)
c0fe08f
Raw
History Blame Contribute Delete
33.8 kB
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8" />
<meta name="viewport" content="width=device-width, initial-scale=1, viewport-fit=cover" />
<meta name="theme-color" content="#F2EDE3" />
<title>pulpie β€” encoder vs decoder</title>
<meta name="color-scheme" content="light" />
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=IBM+Plex+Mono:wght@400;500;600&family=Inter:wght@400;500;600;700&family=Lora:ital@0;1&display=swap" rel="stylesheet">
<style>
:root{
--paper:#F2EDE3; --raised:#FBF7F0; --sunken:#ECE5D7;
--ink:#1B1714; --body-ink:#3F382F; --soft:#756A5E;
--orange:#C6531D; --orange-hover:#E0823A; --pine:#2F5A3C; --hairline:#D7CBB6;
--green:#2F5A3C; --err:#8A2E16; --err-bg:#FFF3EC;
--r:14px; --r-sm:10px; --r-lg:20px;
--sh1:0 1px 2px rgba(27,23,20,.05), 0 1px 3px rgba(27,23,20,.04);
--sh2:0 10px 32px -12px rgba(27,23,20,.18), 0 2px 8px -3px rgba(27,23,20,.07);
--ring:0 0 0 3px rgba(198,83,29,.22);
--safe-t:env(safe-area-inset-top,0px);
color-scheme:light;
}
*{box-sizing:border-box;margin:0;padding:0}
html,body{background:var(--paper);color:var(--ink)}
body{font-family:'Inter','Helvetica Neue',Arial,sans-serif;min-height:100vh;min-height:100dvh;display:flex;flex-direction:column;-webkit-font-smoothing:antialiased;text-rendering:optimizeLegibility;-webkit-tap-highlight-color:transparent}
.mono{font-family:'IBM Plex Mono',monospace}
.wrap{width:100%;max-width:920px;margin:0 auto;padding:max(30px,var(--safe-t)) 20px 28px;flex:1}
::selection{background:rgba(198,83,29,.18)}
*{scrollbar-width:thin;scrollbar-color:var(--hairline) transparent}
::-webkit-scrollbar{width:9px;height:9px}
::-webkit-scrollbar-thumb{background:var(--hairline);border-radius:20px;border:2px solid var(--raised)}
::-webkit-scrollbar-thumb:hover{background:var(--soft)}
::-webkit-scrollbar-track{background:transparent}
a,button,select,textarea{touch-action:manipulation}
/* ── Header ── */
#header-block{margin-bottom:20px}
.topbar{display:flex;justify-content:space-between;align-items:center;gap:12px;margin-bottom:18px}
#eyebrow{font-family:'IBM Plex Mono',monospace;font-size:11px;letter-spacing:.18em;color:var(--orange);text-transform:uppercase;font-weight:500;line-height:1.3}
.pill{display:inline-flex;align-items:center;gap:7px;font-family:'IBM Plex Mono',monospace;font-size:10px;letter-spacing:.14em;text-transform:uppercase;color:var(--soft);background:var(--raised);border:1px solid var(--hairline);padding:6px 11px;border-radius:100px;box-shadow:var(--sh1);transition:all .2s;white-space:nowrap}
.pill .dot{width:7px;height:7px;border-radius:50%;background:var(--green);box-shadow:0 0 0 3px rgba(47,90,60,.15)}
.pill.connecting .dot{background:var(--orange);box-shadow:0 0 0 3px rgba(198,83,29,.15);animation:pulse 1.1s infinite}
.pill.error{color:var(--err);border-color:var(--orange)}
.pill.error .dot{background:var(--err);box-shadow:0 0 0 3px rgba(138,46,22,.15)}
@keyframes pulse{0%,100%{opacity:1;transform:scale(1)}50%{opacity:.4;transform:scale(.85)}}
#headline h1{font-weight:700;letter-spacing:-.035em;font-size:50px;color:var(--ink);line-height:1.02;background:linear-gradient(120deg,var(--ink) 30%,var(--orange) 145%);-webkit-background-clip:text;background-clip:text;-webkit-text-fill-color:transparent}
#lead{font-family:'Lora',serif;font-size:18px;color:var(--body-ink);font-style:italic;max-width:620px;line-height:1.4;margin-top:6px}
/* ── Unified race card ── */
.race-card{background:var(--raised);border:1px solid var(--hairline);border-radius:var(--r-lg);box-shadow:var(--sh2);overflow:hidden;margin-top:20px}
.card-section{padding:18px}
.card-divider{height:1px;background:var(--hairline)}
/* input section */
.input-eyebrow{font-family:'IBM Plex Mono',monospace;font-size:10px;letter-spacing:.16em;color:var(--soft);text-transform:uppercase;margin-bottom:11px;display:flex;align-items:center;gap:8px}
.input-eyebrow::after{content:"";flex:1;height:1px;background:var(--hairline)}
#input-row{display:flex;align-items:stretch;gap:12px}
#example-select{flex:1;height:52px;background:var(--paper);border:1px solid var(--hairline);color:var(--ink);font-family:'IBM Plex Mono',monospace;font-size:13px;padding:0 42px 0 16px;cursor:pointer;appearance:none;border-radius:var(--r-sm);transition:all .15s;background-image:url("data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' width='14' height='14' viewBox='0 0 24 24' fill='none' stroke='%23C6531D' stroke-width='2.5' stroke-linecap='round' stroke-linejoin='round'%3E%3Cpath d='M6 9l6 6 6-6'/%3E%3C/svg%3E");background-repeat:no-repeat;background-position:right 16px center}
#example-select:hover{border-color:var(--orange);box-shadow:var(--sh1)}
#example-select:focus-visible{outline:none;box-shadow:var(--ring);border-color:var(--orange)}
#run-btn{height:52px;min-width:150px;background:linear-gradient(135deg,var(--orange) 0%,#D9642A 100%);color:#FBF7F0;border:none;cursor:pointer;font-family:'IBM Plex Mono',monospace;font-size:13px;font-weight:600;letter-spacing:.1em;text-transform:uppercase;border-radius:var(--r-sm);display:inline-flex;align-items:center;justify-content:center;gap:10px;transition:all .18s;box-shadow:0 4px 14px -4px rgba(198,83,29,.5)}
#run-btn:hover:not(:disabled){filter:brightness(1.06);transform:translateY(-1px);box-shadow:0 7px 22px -5px rgba(198,83,29,.55)}
#run-btn:active:not(:disabled){transform:translateY(0)}
#run-btn:focus-visible{outline:none;box-shadow:var(--ring)}
#run-btn:disabled{opacity:.6;cursor:progress;transform:none}
#run-btn .bolt{width:15px;height:15px}
.spinner{width:15px;height:15px;border:2px solid rgba(251,247,240,.35);border-top-color:#FBF7F0;border-radius:50%;animation:spin .7s linear infinite;display:none}
#run-btn.running .spinner{display:inline-block}
#run-btn.running .bolt{display:none}
@keyframes spin{to{transform:rotate(360deg)}}
#html-accordion{margin-top:12px}
.acc-header{display:flex;align-items:center;gap:9px;padding:7px 2px;cursor:pointer;user-select:none}
.acc-header .chev{font-family:'IBM Plex Mono',monospace;font-size:10px;letter-spacing:.14em;text-transform:uppercase;color:var(--soft);display:flex;align-items:center;gap:7px;transition:color .15s}
.acc-header .chev .arw{display:inline-block;transition:transform .22s;color:var(--orange)}
.acc-header:hover .chev{color:var(--ink)}
#html-accordion.open .chev .arw{transform:rotate(90deg)}
.acc-body{max-height:0;overflow:hidden;transition:max-height .3s ease}
#html-accordion.open .acc-body{max-height:300px}
.acc-body-inner{padding-top:8px}
#custom-html{width:100%;min-height:120px;background:var(--paper);border:1px solid var(--hairline);color:var(--ink);padding:13px 14px;font-family:'IBM Plex Mono',monospace;font-size:12px;line-height:1.55;resize:vertical;border-radius:var(--r-sm);transition:all .15s}
#custom-html:focus-visible{outline:none;border-color:var(--orange);box-shadow:var(--ring)}
.acc-hint{font-family:'Lora',serif;font-style:italic;color:var(--soft);font-size:12px;margin-top:8px}
/* comparison strip */
.cmp-strip{display:grid;grid-template-columns:1fr auto 1fr;gap:14px;align-items:stretch;padding:16px 18px;background:linear-gradient(180deg,var(--raised),rgba(236,229,215,.4))}
.chip{background:var(--paper);border:1px solid var(--hairline);border-radius:var(--r-sm);padding:11px 13px;display:flex;flex-direction:column;gap:8px;transition:all .25s;min-width:0}
.chip.live{border-color:var(--orange);box-shadow:0 0 0 3px rgba(198,83,29,.09)}
.chip.dripper.live{border-color:var(--pine);box-shadow:0 0 0 3px rgba(47,90,60,.09)}
.chip-top{display:flex;align-items:center;justify-content:space-between;gap:8px}
.chip-name{font-family:'IBM Plex Mono',monospace;font-size:10.5px;letter-spacing:.06em;text-transform:uppercase;color:var(--soft);display:flex;align-items:center;gap:7px;min-width:0}
.chip-name b{color:var(--ink);font-weight:600;white-space:nowrap;overflow:hidden;text-overflow:ellipsis}
.chip .led{width:8px;height:8px;border-radius:50%;flex:none;opacity:.4;transition:opacity .25s}
.chip.pulpie .led{background:var(--orange)}
.chip.dripper .led{background:var(--pine)}
.chip.live .led{opacity:1;box-shadow:0 0 0 4px rgba(198,83,29,.12)}
.chip.dripper.live .led{box-shadow:0 0 0 4px rgba(47,90,60,.12)}
.timer{font-family:'IBM Plex Mono',monospace;font-size:13px;letter-spacing:.02em;color:var(--ink);font-variant-numeric:tabular-nums;white-space:nowrap}
.timer .running{color:var(--green);font-weight:500}
.chip.pulpie .timer{color:var(--orange)}
.bar{height:4px;background:var(--sunken);border-radius:4px;position:relative;overflow:hidden}
.bar .fill{position:absolute;inset:0 100% 0 0;background:linear-gradient(90deg,var(--orange),var(--orange-hover));border-radius:4px;transition:right .35s cubic-bezier(.4,0,.2,1)}
.chip.dripper .bar .fill{background:linear-gradient(90deg,var(--pine),#4A7E55)}
.bar .fill.indeterminate{animation:indet 1.3s ease-in-out infinite}
@keyframes indet{0%{left:-30%;right:100%}50%{right:40%}100%{left:100%;right:-30%}}
.cmp-center{display:flex;flex-direction:column;align-items:center;justify-content:center;text-align:center;padding:0 4px;min-width:92px}
.sp-num{font-family:'IBM Plex Mono',monospace;font-weight:700;font-size:32px;line-height:1;letter-spacing:-.03em;color:var(--orange);transition:all .3s}
.sp-lbl{font-family:'IBM Plex Mono',monospace;font-size:9px;letter-spacing:.14em;text-transform:uppercase;color:var(--soft);margin-top:7px;line-height:1.3}
.cmp-center.idle .sp-num{color:var(--hairline);font-size:22px}
.cmp-center.idle .sp-lbl{opacity:0}
/* toolbar (toggle + actions) */
.read-toolbar{display:flex;align-items:center;gap:10px;padding:11px 18px;border-top:1px solid var(--hairline);border-bottom:1px solid var(--hairline);flex-wrap:wrap;background:rgba(236,229,215,.32)}
.toggle-seg{display:flex;gap:3px;background:var(--sunken);border:1px solid var(--hairline);padding:3px;border-radius:100px}
.toggle-seg button{border:none;background:transparent;color:var(--soft);font-family:'IBM Plex Mono',monospace;font-size:10.5px;letter-spacing:.08em;text-transform:uppercase;padding:8px 14px;cursor:pointer;border-radius:100px;transition:all .18s;font-weight:500;display:flex;align-items:center;gap:7px}
.toggle-seg button .led{width:7px;height:7px;border-radius:50%;opacity:.5}
.toggle-seg button[data-model="pulpie"] .led{background:var(--orange)}
.toggle-seg button[data-model="dripper"] .led{background:var(--pine)}
.toggle-seg button.active{background:var(--raised);color:var(--ink);box-shadow:var(--sh1)}
.toggle-seg button.active .led{opacity:1}
.toolbar-spacer{flex:1}
.mini-btn{background:var(--paper);color:var(--ink);border:1px solid var(--hairline);font-family:'IBM Plex Mono',monospace;font-size:10.5px;letter-spacing:.08em;padding:8px 12px;cursor:pointer;transition:all .15s;text-transform:uppercase;border-radius:7px;display:inline-flex;align-items:center;gap:6px;min-height:34px}
.mini-btn:hover{color:var(--orange);border-color:var(--orange);background:var(--raised)}
.mini-btn:focus-visible{outline:none;box-shadow:var(--ring)}
.mini-btn:disabled{opacity:.4;cursor:not-allowed}
/* single reading pane */
.read-pane{min-height:280px;max-height:560px;overflow-y:auto;padding:18px 20px;-webkit-overflow-scrolling:touch}
.read-pane .placeholder{color:var(--soft);font-style:italic;font-family:'Lora',serif;font-size:14.5px;line-height:1.5;display:flex;align-items:center;justify-content:center;min-height:220px;text-align:center;opacity:.85;padding:0 6px}
.sk-line{height:10px;background:linear-gradient(90deg,var(--sunken),var(--hairline),var(--sunken));background-size:200% 100%;animation:shimmer 1.4s infinite;border-radius:6px;margin:14px 0}
.sk-line:nth-child(1){width:55%}.sk-line:nth-child(2){width:88%}.sk-line:nth-child(3){width:72%}.sk-line:nth-child(4){width:90%}.sk-line:nth-child(5){width:40%}
.sk-line:nth-child(2){animation-delay:.1s}.sk-line:nth-child(3){animation-delay:.2s}.sk-line:nth-child(4){animation-delay:.3s}.sk-line:nth-child(5){animation-delay:.4s}
@keyframes shimmer{0%{background-position:200% 0}100%{background-position:-200% 0}}
/* status line (errors / bad input) */
#status-line{font-family:'IBM Plex Mono',monospace;font-size:12px;color:var(--err);padding:0 20px 14px;display:none}
#status-line.show{display:block}
/* markdown */
.md h1,.md h2,.md h3,.md h4,.md h5,.md h6{color:var(--ink);font-weight:700;letter-spacing:-.02em;margin:16px 0 7px;line-height:1.22}
.md h1{font-size:25px}.md h2{font-size:20px}.md h3{font-size:18px}.md h4{font-size:15.5px}
.md h1:first-child,.md h2:first-child{margin-top:0}
.md p{font-family:'Lora',serif;color:var(--body-ink);margin:9px 0;line-height:1.62;font-size:15px}
.md ul,.md ol{margin:9px 0 9px 22px;font-family:'Lora',serif;color:var(--body-ink);font-size:15px}
.md li{margin:4px 0;line-height:1.55}
.md a{color:var(--orange);text-decoration:underline;text-underline-offset:2px;text-decoration-color:var(--hairline)}
.md a:hover{color:var(--orange-hover);text-decoration-color:var(--orange)}
.md code{background:var(--sunken);border:1px solid var(--hairline);padding:1px 6px;font-family:'IBM Plex Mono',monospace;font-size:.86em;color:var(--ink);border-radius:5px}
.md pre{background:var(--sunken);border:1px solid var(--hairline);padding:13px;overflow-x:auto;margin:11px 0;border-radius:var(--r-sm);font-size:12px;line-height:1.5}
.md pre code{background:none;border:none;padding:0;color:var(--ink)}
.md blockquote{border-left:3px solid var(--orange);padding:2px 0 2px 14px;margin:11px 0;color:var(--body-ink);font-style:italic;font-family:'Lora',serif}
.md img{max-width:100%;height:auto;border-radius:var(--r-sm)}
.md table{border-collapse:collapse;margin:11px 0;font-family:'IBM Plex Mono',monospace;font-size:12.5px;width:100%;display:block;overflow-x:auto}
.md th,.md td{border:1px solid var(--hairline);padding:7px 10px;text-align:left;white-space:nowrap}
.md th{background:var(--sunken)}
.md hr{border:none;border-top:1px solid var(--hairline);margin:15px 0}
/* flash */
.flash{background:var(--err-bg);color:var(--err);border:1px solid var(--orange);padding:12px 16px;font-family:'IBM Plex Mono',monospace;font-size:12.5px;margin-top:14px;border-radius:var(--r-sm);display:none;align-items:center;gap:10px}
.flash.show{display:flex}
.flash .x{margin-left:auto;cursor:pointer;opacity:.6;padding:2px 6px}
.flash .x:hover{opacity:1}
/* ── Bench ── */
#bench-well{background:var(--raised);border:1px solid var(--hairline);border-radius:var(--r);padding:20px;margin-top:18px;box-shadow:var(--sh1)}
#bench-well .label{font-family:'IBM Plex Mono',monospace;font-size:10.5px;letter-spacing:.14em;color:var(--soft);margin-bottom:14px;text-transform:uppercase;display:flex;align-items:center;gap:8px}
#bench-well .label::after{content:"";flex:1;height:1px;background:var(--hairline)}
.bench-row{display:grid;grid-template-columns:170px 1fr 80px;align-items:center;gap:14px;padding:8px 0}
.bench-row .who{font-family:'Inter',sans-serif;font-size:13.5px;color:var(--ink);font-weight:500;white-space:nowrap;overflow:hidden;text-overflow:ellipsis}
.bench-row .who .params{color:var(--soft);font-family:'IBM Plex Mono',monospace;font-size:11px;margin-left:6px;font-weight:400}
.bench-track{height:10px;background:var(--sunken);border-radius:6px;overflow:hidden}
.bench-fill{height:100%;border-radius:6px;background:var(--soft);width:0;transition:width .9s cubic-bezier(.4,0,.2,1)}
.bench-row.win .bench-fill{background:linear-gradient(90deg,var(--orange),var(--orange-hover))}
.bench-row.dim .bench-fill{opacity:.5}
.bench-row .score{font-family:'IBM Plex Mono',monospace;font-size:13px;text-align:right;font-variant-numeric:tabular-nums;color:var(--ink)}
.bench-row.win .score{color:var(--orange);font-weight:600}
#bench-well .note{font-family:'Lora',serif;font-style:italic;color:var(--body-ink);margin-top:12px;font-size:13.5px;line-height:1.5}
/* footer */
#footer-mono{font-family:'IBM Plex Mono',monospace;font-size:11.5px;letter-spacing:.06em;color:var(--soft);border-top:1px solid var(--hairline);padding-top:16px;margin-top:24px;padding-bottom:8px;display:flex;flex-wrap:wrap;gap:3px 0;align-items:center}
#footer-mono a{color:var(--ink);text-decoration:none;padding:3px;transition:color .15s}
#footer-mono a:hover{color:var(--orange)}
#footer-mono .sep{opacity:.4;margin:0 5px}
/* ── Mobile ── */
@media (max-width:600px){
.wrap{padding:max(20px,var(--safe-t)) 14px 22px}
#headline h1{font-size:38px}
#lead{font-size:16px}
.race-card{border-radius:var(--r)}
.card-section{padding:15px}
#input-row{flex-direction:column}
#example-select,#run-btn{flex:1;width:100%;min-width:0}
/* comparison strip stacks: pulpie / speedup / dripper */
.cmp-strip{grid-template-columns:1fr;gap:10px;padding:14px}
.cmp-center{flex-direction:row;gap:12px;min-width:0;padding:2px 0;justify-content:flex-start}
.cmp-center .sp-num{font-size:24px}
.cmp-center .sp-lbl{margin-top:0}
.chip{flex-direction:row;align-items:center;gap:14px;flex-wrap:wrap}
.chip-top{flex:1;min-width:120px}
.chip .bar{flex:1;min-width:80px}
.read-toolbar{padding:10px 14px}
.read-pane{padding:16px 16px;min-height:240px}
.bench-row{grid-template-columns:1fr 60px;gap:6px 12px}
.bench-row .who{grid-column:1/3}
.bench-row .bench-track{grid-column:1}
.bench-row .score{grid-column:2}
}
@media (max-width:420px){
#eyebrow{font-size:9.5px}
.pill{font-size:9px;padding:5px 9px}
#headline h1{font-size:32px}
.sp-num{font-size:22px}
}
@media (prefers-reduced-motion:reduce){*,*::before,*::after{animation-duration:.01ms!important;animation-iteration-count:1!important;transition-duration:.01ms!important}}
</style>
</head>
<body>
<div class="wrap">
<!-- Header -->
<div id="header-block">
<div class="topbar">
<div id="eyebrow">FEYN Β· CONTENT EXTRACTION Β· ENCODER VS DECODER</div>
<div class="pill connecting" id="conn-pill"><span class="dot"></span><span id="conn-text">CONNECTING</span></div>
</div>
<div id="headline"><h1>pulpie</h1></div>
<div id="lead">Extract main content from HTML. One forward pass, not one token at a time.</div>
</div>
<!-- Unified race card -->
<div class="race-card">
<!-- input -->
<div class="card-section">
<div class="input-eyebrow">EXAMPLE PAGE</div>
<div id="input-row">
<select id="example-select" aria-label="Example page">
<option value="paulgraham.com">paulgraham.com</option>
<option value="nodejs.dev">nodejs.dev</option>
<option value="protobuf.dev">protobuf.dev</option>
<option value="istio.io">istio.io</option>
<option value="tutorialspoint.com">tutorialspoint.com</option>
<option value="builtin.com">builtin.com</option>
<option value="bbc.com">bbc.com</option>
<option value="theguardian.com">theguardian.com</option>
<option value="wikipedia.org">wikipedia.org</option>
<option value="stackoverflow.com">stackoverflow.com</option>
</select>
<button id="run-btn">
<svg class="bolt" viewBox="0 0 24 24" fill="currentColor"><path d="M13 2L4.5 13.5H11l-1 8.5L19.5 10H13z"/></svg>
<span class="spinner"></span>
<span class="btn-label">EXTRACT β†’</span>
</button>
</div>
<div id="html-accordion">
<div class="acc-header" id="acc-toggle" role="button" tabindex="0" aria-expanded="false">
<span class="chev"><span class="arw">β–Έ</span>OR PASTE RAW HTML</span>
</div>
<div class="acc-body"><div class="acc-body-inner">
<textarea id="custom-html" placeholder="<html>…</html> (2 MB max; JS-rendered SPAs will come up empty β€” nothing static to extract)"></textarea>
<div class="acc-hint">If set, this overrides the example dropdown. ⌘/Ctrl+Enter to run.</div>
</div></div>
</div>
</div>
<div class="card-divider"></div>
<!-- comparison strip (both models, always visible) -->
<div class="cmp-strip">
<div class="chip pulpie" id="pulpie-chip">
<div class="chip-top">
<div class="chip-name"><span class="led"></span>PULPIE <b>210M Β· ENCODER</b></div>
<div class="timer" id="pulpie-timer"></div>
</div>
<div class="bar"><div class="fill" id="pulpie-bar"></div></div>
</div>
<div class="cmp-center idle" id="cmp-center">
<div class="sp-num" id="sp-num">β€”</div>
<div class="sp-lbl" id="sp-lbl">Γ— FASTER Β· ON THIS PAGE</div>
</div>
<div class="chip dripper" id="dripper-chip">
<div class="chip-top">
<div class="chip-name"><span class="led"></span>DRIPPER <b>0.6B Β· DECODER</b></div>
<div class="timer" id="dripper-timer"></div>
</div>
<div class="bar"><div class="fill" id="dripper-bar"></div></div>
</div>
</div>
<!-- toolbar: toggle + actions -->
<div class="read-toolbar">
<div class="toggle-seg" role="tablist" aria-label="Which model to read">
<button data-model="pulpie" class="active" role="tab" aria-selected="true"><span class="led"></span>PULPIE</button>
<button data-model="dripper" role="tab" aria-selected="false"><span class="led"></span>DRIPPER</button>
</div>
<div class="toolbar-spacer"></div>
<button class="mini-btn" id="copy-btn">⧉ COPY MD</button>
<button class="mini-btn" id="diff-btn">β‰  DIFF</button>
</div>
<!-- single reading pane -->
<div class="read-pane" id="read-pane">
<div class="placeholder">One encoder forward pass over every block.</div>
</div>
<div id="status-line"></div>
</div>
<div class="flash" id="flash"><span id="flash-text"></span><span class="x" id="flash-x">βœ•</span></div>
<!-- Bench -->
<div id="bench-well">
<div class="label">WEBMAINBENCH Β· 6,647 PAGES Β· ROUGE-5 F1</div>
<div class="bench-row win"><div class="who">pulpie orange-small<span class="params">210M</span></div><div class="bench-track"><div class="bench-fill" data-pct="86.2"></div></div><div class="score">0.862</div></div>
<div class="bench-row"><div class="who">dripper (MinerU-HTML)<span class="params">0.6B</span></div><div class="bench-track"><div class="bench-fill" data-pct="86.4"></div></div><div class="score">0.864</div></div>
<div class="bench-row dim"><div class="who">trafilatura<span class="params">β€”</span></div><div class="bench-track"><div class="bench-fill" data-pct="61.9"></div></div><div class="score">0.619</div></div>
<div class="note">Same quality. ~$7,900 to clean 1B pages, versus ~$159,000. <span style="color:var(--soft);font-family:'IBM Plex Mono',monospace;font-style:normal;font-size:11px;letter-spacing:.05em">β€” 13.7 pg/s on L4 vs 0.68 pg/s</span></div>
</div>
<!-- Footer -->
<div id="footer-mono">
<span class="sep">pip install pulpie</span><span class="sep">Β·</span>
<a href="https://github.com/feyninc/pulpie">GITHUB</a><span class="sep">Β·</span>
<a href="https://huggingface.co/feyninc/pulpie-orange-small">MODEL</a><span class="sep">Β·</span>
<a href="https://github.com/feyninc/pulpie/blob/main/BENCHMARKS.md">BENCHMARKS</a><span class="sep">Β·</span>
<a href="https://usefeyn.com/blog/pulpie-pareto-optimal-models-for-cleaning-the-web/">BLOG</a><span class="sep">Β·</span>
BUILT BY <a href="https://usefeyn.com">FEYN</a>
</div>
</div>
<script type="module">
import { Client } from "https://cdn.jsdelivr.net/npm/@gradio/client/dist/index.min.js";
import { marked } from "https://cdn.jsdelivr.net/npm/marked@12/lib/marked.esm.js";
import DOMPurify from "https://cdn.jsdelivr.net/npm/dompurify@3/dist/purify.es.mjs";
marked.setOptions({ gfm:true, breaks:false, headerIds:false, mangle:false });
const $ = (id) => document.getElementById(id);
const runBtn=$("run-btn"), btnLabel=runBtn.querySelector(".btn-label");
const exampleSelect=$("example-select"), customHtml=$("custom-html");
const readPane=$("read-pane");
const pulpieChip=$("pulpie-chip"), dripperChip=$("dripper-chip");
const pulpieTimer=$("pulpie-timer"), dripperTimer=$("dripper-timer");
const pulpieBar=$("pulpie-bar"), dripperBar=$("dripper-bar");
const cmpCenter=$("cmp-center"), spNum=$("sp-num"), spLbl=$("sp-lbl");
const statusLine=$("status-line");
const flashEl=$("flash"), flashText=$("flash-text");
const connPill=$("conn-pill"), connText=$("conn-text");
const copyBtn=$("copy-btn"), diffBtn=$("diff-btn");
const PLACEHOLDERS = {
pulpie:'<div class="placeholder">One encoder forward pass over every block.</div>',
dripper:'<div class="placeholder">The same labels, one token at a time.</div>',
};
const SKEL = '<div class="sk-line"></div><div class="sk-line"></div><div class="sk-line"></div><div class="sk-line"></div><div class="sk-line"></div>';
let lastMd = { pulpie:"", dripper:"" };
let running = false, currentModel = "pulpie";
function renderMarkdown(md){ if(!md) return ""; return DOMPurify.sanitize(marked.parse(md)); }
function renderRead(){
const md = lastMd[currentModel];
if(md) readPane.innerHTML = renderMarkdown(md);
else if(running) readPane.innerHTML = SKEL;
else readPane.innerHTML = PLACEHOLDERS[currentModel];
}
function setRunning(on){
running = on;
runBtn.classList.toggle("running", on); runBtn.disabled=on;
btnLabel.textContent = on ? "EXTRACTING…" : "EXTRACT β†’";
if(on || lastMd[currentModel]) renderRead();
}
function showError(msg){ flashText.textContent=msg; flashEl.classList.add("show"); }
function clearError(){ flashEl.classList.remove("show"); }
function setStatus(html){ if(!html){ statusLine.classList.remove("show"); statusLine.innerHTML=""; return; } statusLine.innerHTML=html; statusLine.classList.add("show"); }
function setBar(bar, pct, indeterminate){
bar.classList.toggle("indeterminate", !!indeterminate);
if(indeterminate){ bar.style.right=""; }
else { bar.style.right = (100-Math.max(0,Math.min(100,pct)))+"%"; }
}
function setSpeedupIdle(){ cmpCenter.classList.add("idle"); spNum.textContent="β€”"; spLbl.textContent="Γ— FASTER Β· ON THIS PAGE"; }
function setSpeedup(num){
cmpCenter.classList.remove("idle");
spNum.textContent = num;
spLbl.textContent = "Γ— FASTER Β· ON THIS PAGE";
}
// Pull the speedup token ("12x" / "9.4x") out of the backend's verdict HTML:
// <div ...>PULPIE 0.42s β†’ DRIPPER 2.8s Β· <span ...>12x ON THIS PAGE</span></div>
function applyVerdict(v){
if(!v){ return; }
const m = v.match(/<span[^>]*>\s*([\d.]+x)\s+ON THIS PAGE\s*<\/span>/);
if(m){ setSpeedup(m[1]); setStatus(""); }
else { setSpeedupIdle(); setStatus(v); } // plain-text verdict (bad input / error)
}
function resetPanes(){
lastMd={pulpie:"",dripper:""};
pulpieChip.classList.remove("live"); dripperChip.classList.remove("live");
pulpieTimer.innerHTML=""; dripperTimer.innerHTML="";
setBar(pulpieBar,0); setBar(dripperBar,0);
setSpeedupIdle(); setStatus(""); clearError();
renderRead();
}
function normalizeRecord(d){
if(!d) return null;
if(typeof d==="object" && !Array.isArray(d) && "pulpie_md" in d) return d;
if(Array.isArray(d)){ for(const it of d){ if(it && typeof it==="object" && "pulpie_md" in it) return it; } }
return null;
}
function applyRecord(rec){
if(!rec) return;
if(rec.pulpie_md!==undefined && rec.pulpie_md!==""){
lastMd.pulpie=rec.pulpie_md;
pulpieChip.classList.add("live");
setBar(pulpieBar,100);
setTimeout(()=>pulpieChip.classList.remove("live"),900);
if(currentModel==="pulpie") renderRead();
}
if(rec.dripper_md!==undefined && rec.dripper_md!==""){
lastMd.dripper=rec.dripper_md;
dripperChip.classList.add("live");
if(currentModel==="dripper") renderRead();
}
if(rec.pulpie_timer!==undefined) pulpieTimer.innerHTML=rec.pulpie_timer;
if(rec.dripper_timer!==undefined) dripperTimer.innerHTML=rec.dripper_timer;
// dripper progress: indeterminate while streaming, full when the verdict lands
if(rec.dripper_md && !(rec.verdict)){ setBar(dripperBar,0,true); }
if(rec.verdict!==undefined && rec.verdict!==""){
applyVerdict(rec.verdict);
dripperChip.classList.remove("live");
setBar(dripperBar,100);
}
}
// ── Connect ──
let client=null;
try {
client = await Client.connect(window.location.origin, { events:["data","status"] });
connPill.classList.remove("connecting"); connText.textContent="READY";
} catch(e){
connPill.classList.remove("connecting"); connPill.classList.add("error"); connText.textContent="OFFLINE";
showError("Could not connect to backend: " + (e?.message||e));
}
// ── Run ──
async function run(){
if(!client){ showError("Backend not connected. Reload the page."); return; }
resetPanes(); setRunning(true);
const params = { example_name: exampleSelect.value, custom_html: customHtml.value.trim() };
try {
const submission = client.submit("/race", params);
let sawData=false;
for await (const msg of submission){
if(msg.type==="status"){
if(msg.stage==="pending" && msg.position!=null){
setStatus('<span style="color:var(--soft)">QUEUED Β· POSITION '+msg.position+'</span>');
} else if(msg.stage==="generating"){ if(!sawData) setStatus(""); }
continue;
}
if(msg.type==="data"){ sawData=true; applyRecord(normalizeRecord(msg.data)); }
}
if(!sawData) showError("Backend returned no data. Check the /race endpoint and Space logs.");
} catch(e){ showError("Race failed: " + (e?.message||e)); }
finally { setRunning(false); }
}
runBtn.addEventListener("click", run);
exampleSelect.addEventListener("keydown", e=>{ if(e.key==="Enter"){ e.preventDefault(); run(); } });
customHtml.addEventListener("keydown", e=>{ if(e.key==="Enter" && (e.metaKey||e.ctrlKey)){ e.preventDefault(); run(); } });
// ── Accordion ──
const acc=$("html-accordion"), accToggle=$("acc-toggle");
accToggle.addEventListener("click", ()=>{ const o=acc.classList.toggle("open"); accToggle.setAttribute("aria-expanded", o); });
accToggle.addEventListener("keydown", e=>{ if(e.key==="Enter"||e.key===" "){ e.preventDefault(); accToggle.click(); } });
// ── Model toggle ──
document.querySelectorAll(".toggle-seg button").forEach(b=>{
b.addEventListener("click", ()=>{
currentModel = b.getAttribute("data-model");
document.querySelectorAll(".toggle-seg button").forEach(x=>{ const on=x===b; x.classList.toggle("active",on); x.setAttribute("aria-selected",on); });
renderRead();
});
});
// ── Flash dismiss ──
$("flash-x").addEventListener("click", clearError);
// ── Copy ──
copyBtn.addEventListener("click", async ()=>{
const text = lastMd[currentModel];
if(!text) return;
try { await navigator.clipboard.writeText(text); copyBtn.textContent="βœ“ COPIED"; }
catch { copyBtn.textContent="COPY FAILED"; }
setTimeout(()=>(copyBtn.textContent="⧉ COPY MD"), 1400);
});
// ── Diff (against the other model, shown in the single reading pane) ──
let diffOpen=false, diffForModel=null;
diffBtn.addEventListener("click", ()=>{
if(diffOpen && diffForModel===currentModel){ renderRead(); diffOpen=false; diffBtn.textContent="β‰  DIFF"; return; }
const other = currentModel==="pulpie" ? "dripper" : "pulpie";
if(!lastMd[currentModel] || !lastMd[other]) return;
const a=lastMd.pulpie, b=lastMd.dripper, same=a.trim()===b.trim();
if(same){
readPane.innerHTML = renderMarkdown(lastMd[currentModel]) +
'<div style="font-family:\'IBM Plex Mono\',monospace;font-size:11px;color:var(--green);margin-top:14px;border-top:1px solid var(--hairline);padding-top:10px;">βœ“ IDENTICAL OUTPUTS β€” PULPIE &amp; DRIPPER MATCH</div>';
diffBtn.textContent="IDENTICAL";
} else {
readPane.innerHTML = '<pre style="white-space:pre-wrap;font-family:\'IBM Plex Mono\',monospace;font-size:12px;line-height:1.5;background:var(--sunken);border:1px solid var(--hairline);padding:12px;border-radius:8px;">'+escapeHtml(lineDiff(a,b))+'</pre>';
diffBtn.textContent="↩ BACK";
}
diffOpen=true; diffForModel=currentModel;
});
function lineDiff(a,b){
const la=a.split("\n"), lb=b.split("\n"), n=la.length, m=lb.length;
const dp=Array.from({length:n+1}, ()=>new Array(m+1).fill(0));
for(let i=n-1;i>=0;i--) for(let j=m-1;j>=0;j--) dp[i][j]=la[i]===lb[j]?dp[i+1][j+1]+1:Math.max(dp[i+1][j],dp[i][j+1]);
let i=0,j=0,out=[];
while(i<n&&j<m){ if(la[i]===lb[j]){ out.push(" "+la[i]); i++; j++; } else if(dp[i+1][j]>=dp[i][j+1]){ out.push("- "+la[i]); i++; } else { out.push("+ "+lb[j]); j++; } }
while(i<n) out.push("- "+la[i++]); while(j<m) out.push("+ "+lb[j++]);
return out.slice(-400).join("\n");
}
function escapeHtml(s){ return s.replace(/&/g,"&amp;").replace(/</g,"&lt;").replace(/>/g,"&gt;"); }
// ── Animate benchmark bars on first paint ──
requestAnimationFrame(()=>{ setTimeout(()=>{
requestAnimationFrame(()=>{ document.querySelectorAll(".bench-fill").forEach(f=>{ f.style.width=(f.getAttribute("data-pct")||0)+"%"; }); });
}, 120); });
</script>
</body>
</html>