Spaces:
Running
Running
Download index.html from srish327/skill-extractor: direct link, hf CLI and curl.
- Browser
- Download file 8.14 kB
-
https://huggingface.co/spaces/srish327/skill-extractor/resolve/main/index.html
- Command line
-
hf download hf://spaces/srish327/skill-extractor/index.html
-
curl -L -o index.html https://huggingface.co/spaces/srish327/skill-extractor/resolve/main/index.html
8.14 kB
| <html lang="en"> | |
| <head> | |
| <meta charset="utf-8"> | |
| <meta name="viewport" content="width=device-width, initial-scale=1"> | |
| <title>Skill Extractor</title> | |
| <style> | |
| :root { | |
| --bg: #fafaf9; --panel: #ffffff; --text: #1c1917; --muted: #78716c; --border: #e7e5e4; | |
| --skill: #15803d; --skill-bg: #dcfce7; --know: #1d4ed8; --know-bg: #dbeafe; --accent: #1c1917; | |
| } | |
| @media (prefers-color-scheme: dark) { | |
| :root:not([data-theme="light"]) { | |
| --bg: #0c0a09; --panel: #1c1917; --text: #f5f5f4; --muted: #a8a29e; --border: #292524; | |
| --skill: #86efac; --skill-bg: #14532d; --know: #93c5fd; --know-bg: #1e3a8a; --accent: #f5f5f4; | |
| } | |
| } | |
| * { box-sizing: border-box; } | |
| body { margin: 0; background: var(--bg); color: var(--text); | |
| font: 15px/1.55 ui-sans-serif, system-ui, -apple-system, "Segoe UI", sans-serif; } | |
| main { max-width: 860px; margin: 0 auto; padding: 32px 16px 64px; } | |
| h1 { font-size: 26px; margin: 0 0 4px; letter-spacing: -0.01em; } | |
| .sub { color: var(--muted); margin: 0 0 24px; } | |
| .sub a { color: inherit; } | |
| textarea { width: 100%; min-height: 170px; padding: 12px; border: 1px solid var(--border); border-radius: 10px; | |
| background: var(--panel); color: var(--text); font: inherit; resize: vertical; } | |
| .row { display: flex; gap: 12px; align-items: center; margin: 12px 0 24px; flex-wrap: wrap; } | |
| button { background: var(--accent); color: var(--bg); border: 0; border-radius: 8px; padding: 10px 18px; | |
| font: inherit; font-weight: 600; cursor: pointer; } | |
| button:disabled { opacity: .45; cursor: default; } | |
| #status { color: var(--muted); font-size: 14px; } | |
| .bar { height: 4px; background: var(--border); border-radius: 2px; overflow: hidden; flex: 1 1 160px; max-width: 260px; } | |
| .bar > div { height: 100%; width: 0; background: var(--accent); transition: width .2s; } | |
| .card { background: var(--panel); border: 1px solid var(--border); border-radius: 10px; padding: 16px; margin-bottom: 16px; } | |
| .card h2 { font-size: 13px; text-transform: uppercase; letter-spacing: .06em; color: var(--muted); margin: 0 0 10px; } | |
| .hl { white-space: pre-wrap; } | |
| mark { border-radius: 4px; padding: 1px 2px; } | |
| mark.SKILL { background: var(--skill-bg); color: var(--skill); } | |
| mark.KNOWLEDGE { background: var(--know-bg); color: var(--know); } | |
| .chips { display: flex; flex-wrap: wrap; gap: 6px; } | |
| .chip { border-radius: 999px; padding: 3px 10px; font-size: 14px; } | |
| .chip.SKILL { background: var(--skill-bg); color: var(--skill); } | |
| .chip.KNOWLEDGE { background: var(--know-bg); color: var(--know); } | |
| .grid { display: grid; grid-template-columns: 1fr 1fr; gap: 16px; } | |
| @media (max-width: 640px) { .grid { grid-template-columns: 1fr; } } | |
| .empty { color: var(--muted); font-size: 14px; } | |
| footer { color: var(--muted); font-size: 13px; margin-top: 32px; } | |
| </style> | |
| </head> | |
| <body> | |
| <main> | |
| <h1>Job Posting Skill Extractor</h1> | |
| <p class="sub"> | |
| Finds <b style="color:var(--skill)">skills</b> (what you can do) and | |
| <b style="color:var(--know)">knowledge</b> (what you know) in a job posting. The model is DeBERTa-v3 with two BIO heads and a learned CRF, | |
| fine-tuned on SkillSpan (test span F1 0.645 ± 0.005). | |
| <b>It runs entirely in your browser</b>, so your text never leaves this page. | |
| <a href="https://huggingface.co/srish327/deberta-v3-skillspan-crf">Model</a> · | |
| <a href="https://github.com/MSR327/skillspan-ner">Code</a> | |
| </p> | |
| <textarea id="text">We are looking for a Data Engineer with strong Python and SQL experience. You will design ETL pipelines on AWS, collaborate with stakeholders, and mentor junior engineers. | |
| Requirements: | |
| - 3+ years of experience with Kubernetes and Terraform | |
| - Excellent communication skills | |
| - Ability to prioritize tasks in a fast-paced environment | |
| Nice to have: Kafka, dbt</textarea> | |
| <div class="row"> | |
| <button id="go" disabled>Extract</button> | |
| <div class="bar"><div id="prog"></div></div> | |
| <span id="status">Loading model (243 MB, cached after the first visit)…</span> | |
| </div> | |
| <div class="grid"> | |
| <div class="card"><h2>Skills</h2><div id="skills" class="chips"></div></div> | |
| <div class="card"><h2>Knowledge</h2><div id="knowledge" class="chips"></div></div> | |
| </div> | |
| <div class="card"><h2>Skills in context</h2><div id="hl-skill" class="hl empty">—</div></div> | |
| <div class="card"><h2>Knowledge in context</h2><div id="hl-know" class="hl empty">—</div></div> | |
| <footer>Skill and knowledge spans can overlap, so each is shown separately. Model is int8-quantized ONNX, run with onnxruntime-web. | |
| Not intended for automated hiring decisions.</footer> | |
| </main> | |
| <script src="https://cdn.jsdelivr.net/npm/onnxruntime-web@1.20.1/dist/ort.min.js"></script> | |
| <script type="module"> | |
| import { AutoTokenizer } from "https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.5.1"; | |
| import { extract } from "./extract.js"; | |
| const REPO = "srish327/deberta-v3-skillspan-crf"; | |
| const BASE = `https://huggingface.co/${REPO}/resolve/main/`; | |
| const MODEL_URL = BASE + "onnx/model_quantized.onnx"; | |
| const $ = (id) => document.getElementById(id); | |
| const status = (s) => ($("status").textContent = s); | |
| async function fetchWithProgress(url) { | |
| let cache = null; | |
| try { cache = await caches.open("skill-extractor-v1"); const hit = await cache.match(url); | |
| if (hit) { $("prog").style.width = "100%"; return await hit.arrayBuffer(); } } catch {} | |
| const res = await fetch(url); | |
| if (!res.ok) throw new Error(`download failed: ${res.status}`); | |
| const total = +res.headers.get("content-length") || 0, reader = res.body.getReader(), parts = []; | |
| let got = 0; | |
| for (;;) { | |
| const { done, value } = await reader.read(); if (done) break; | |
| parts.push(value); got += value.length; | |
| if (total) { $("prog").style.width = `${(100 * got) / total}%`; | |
| status(`Loading model… ${Math.round(got / 1e6)} / ${Math.round(total / 1e6)} MB`); } | |
| } | |
| const buf = await new Blob(parts).arrayBuffer(); | |
| try { await cache?.put(url, new Response(buf.slice(0))); } catch {} | |
| return buf; | |
| } | |
| let tok, session, crfs; | |
| async function load() { | |
| ort.env.wasm.wasmPaths = "https://cdn.jsdelivr.net/npm/onnxruntime-web@1.20.1/dist/"; | |
| [tok, crfs] = await Promise.all([ | |
| AutoTokenizer.from_pretrained(REPO), | |
| fetch(BASE + "skillex_crf.json").then((r) => r.json()), | |
| ]); | |
| const buf = await fetchWithProgress(MODEL_URL); | |
| status("Starting model…"); | |
| session = await ort.InferenceSession.create(buf, { executionProviders: ["wasm"] }); | |
| status("Ready. Paste a job posting and click Extract."); | |
| $("go").disabled = false; | |
| } | |
| const esc = (s) => s.replace(/[&<>"]/g, (c) => ({ "&": "&", "<": "<", ">": ">", '"': """ })[c]); | |
| function highlight(text, spans, type) { | |
| let html = "", pos = 0; | |
| for (const s of spans.filter((x) => x.type === type).sort((a, b) => a.start - b.start)) { | |
| html += esc(text.slice(pos, s.start)) + `<mark class="${type}">${esc(text.slice(s.start, s.end))}</mark>`; | |
| pos = s.end; | |
| } | |
| return html + esc(text.slice(pos)); | |
| } | |
| function chips(el, text, spans, type) { | |
| const uniq = [...new Set(spans.filter((s) => s.type === type).map((s) => text.slice(s.start, s.end)))]; | |
| el.innerHTML = uniq.length ? uniq.map((u) => `<span class="chip ${type}">${esc(u)}</span>`).join("") | |
| : `<span class="empty">none found</span>`; | |
| } | |
| $("go").onclick = async () => { | |
| const text = $("text").value; $("go").disabled = true; status("Running…"); | |
| try { | |
| const t0 = performance.now(), spans = await extract({ ort, tok, session, crfs }, text); | |
| chips($("skills"), text, spans, "SKILL"); chips($("knowledge"), text, spans, "KNOWLEDGE"); | |
| for (const [id, type] of [["hl-skill", "SKILL"], ["hl-know", "KNOWLEDGE"]]) { | |
| $(id).classList.remove("empty"); $(id).innerHTML = highlight(text, spans, type); | |
| } | |
| status(`Done in ${Math.round(performance.now() - t0)} ms, ${spans.length} spans.`); | |
| } catch (e) { status("Error: " + e.message); console.error(e); } | |
| $("go").disabled = false; | |
| }; | |
| load().catch((e) => { status("Failed to load model: " + e.message); console.error(e); }); | |
| </script> | |
| </body> | |
| </html> | |