bf2f59e292
Built from main 4ae9c5ffef
Merge branch 'refactor/task-21-batch1-corrections'
23 lines
40 KiB
HTML
23 lines
40 KiB
HTML
<!DOCTYPE html><html lang="en"> <head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1"><title>AI For Dummies — Browser model lab</title><meta name="description" content="Everything else in this guide talks about models you rent. Here you download one: a small open model from Hugging Face, running on your own GPU or CPU, with no account, no key, and no server reading your prompt."><link rel="canonical" href="https://netcracker.pages.marcospaulo.dev.br/ai-for-dummies/lab/"><link rel="icon" type="image/svg+xml" href="/ai-for-dummies/favicon.svg"><link rel="stylesheet" href="/ai-for-dummies/_astro/tokens.BxNFvepo.css"><link rel="stylesheet" href="/ai-for-dummies/fonts/fonts.css"><link rel="stylesheet" href="/ai-for-dummies/_astro/base.u_-OWNC4.css"><link rel="stylesheet" href="/ai-for-dummies/_astro/chapters.DVZ6ptlR.css"><link rel="stylesheet" href="/ai-for-dummies/_astro/motion.CtPv1C_R.css"><link rel="stylesheet" href="/ai-for-dummies/_astro/lab@_@astro.CFKNgTKZ.css">
|
||
<link rel="stylesheet" href="/ai-for-dummies/_astro/ChapterLayout.BpAlYsUl.css">
|
||
<link rel="stylesheet" href="/ai-for-dummies/_astro/ChapterHero.Bf5I54SK.css">
|
||
<link rel="stylesheet" href="/ai-for-dummies/_astro/FilmPlayer.B1nrpPmm.css">
|
||
<link rel="stylesheet" href="/ai-for-dummies/_astro/require-field.D8D5Or_Q.css"><script type="module" src="/ai-for-dummies/_astro/page.DGRhADao.js"></script></head> <body> <header class="top" id="top" data-astro-cid-xattfbdu> <div class="cell" data-astro-cid-xattfbdu><a href="/ai-for-dummies/summary/" data-astro-cid-t3nv4xef>← ROUTE MAP</a></div> <div class="cell" data-astro-cid-xattfbdu><span data-astro-cid-t3nv4xef>08 / LAB</span></div> <div class="cell" data-astro-cid-xattfbdu><div class="lang-switch" aria-label="Language" data-language-toggle> <button class="active" data-lang="en" aria-pressed="true">EN</button> <span>/</span> <button data-lang="pt" aria-pressed="false">PT</button> </div> <script type="module">function o(t){const n=t.querySelectorAll("[data-lang]");function l(e){document.documentElement.lang=e==="pt"?"pt-BR":"en",n.forEach(a=>{const u=a.dataset.lang===e;a.classList.toggle("active",u),a.setAttribute("aria-pressed",String(u))}),document.querySelectorAll("[data-language-content]").forEach(a=>{a.hidden=a.dataset.languageContent!==e}),window.dispatchEvent(new CustomEvent("ai-for-dummies:languagechange",{detail:{language:e}}))}function g(e){return e==="pt"?"pt":"en"}n.forEach(e=>{e.addEventListener("click",()=>{const a=g(e.dataset.lang);l(a);try{localStorage.setItem("ai-for-dummies-language",a)}catch{}})});let c="en";try{c=localStorage.getItem("ai-for-dummies-language")||"en"}catch{}l(g(c))}document.querySelectorAll("[data-language-toggle]").forEach(t=>{"requestIdleCallback"in window?window.requestIdleCallback(()=>o(t)):setTimeout(()=>o(t),0)});</script></div> </header> <main> <section class="hero rise-in has-aside" data-astro-cid-7xzskqga> <div class="hero-copy" data-astro-cid-7xzskqga> <div data-language-content="en" data-astro-cid-7xzskqga><p data-astro-cid-4yr5atew="true" class="eyebrow tone-red">Browser model lab</p> </div> <div data-language-content="pt" hidden data-astro-cid-7xzskqga><p data-astro-cid-4yr5atew="true" class="eyebrow tone-red">Laboratório de modelo no navegador</p> </div> <h1 class="parallax-near" data-astro-cid-7xzskqga><span data-language-content="en" data-astro-cid-t3nv4xef>Run a model<br><em>in this tab.</em></span><span data-language-content="pt" hidden data-astro-cid-t3nv4xef>Rode um modelo<br><em>nesta aba.</em></span></h1> <div class="intro parallax-far" data-astro-cid-7xzskqga> <p data-language-content="en" data-astro-cid-t3nv4xef>Everything else in this guide talks about models you rent. Here you download one: a small open model from Hugging Face, running on your own GPU or CPU, with no account, no key, and no server reading your prompt.</p> <p data-language-content="pt" hidden data-astro-cid-t3nv4xef>O resto deste guia fala de modelos que você aluga. Aqui você baixa um: um modelo aberto pequeno do Hugging Face, rodando na sua GPU ou CPU, sem conta, sem chave e sem servidor lendo seu prompt.</p> </div> </div> <div class="hero-aside rise-in-late" data-astro-cid-7xzskqga> <figure class="film film-hero" data-film data-paused="true" data-astro-cid-5bv36nto> <div class="frame" data-astro-cid-5bv36nto> <video data-language-content="en" src="/ai-for-dummies/videos/browser-en.mp4" poster="/ai-for-dummies/videos/browser-en.jpg" width="1280" height="720" muted loop playsinline preload="none" aria-label="08 / Browser model. A language model, inside a web page. The weights download once, from the Hub. Next visit, they load from the browser cache. It runs on your own GPU. Your prompt never leaves the page. The model comes to you." data-astro-cid-5bv36nto></video><video data-language-content="pt" hidden src="/ai-for-dummies/videos/browser-pt.mp4" poster="/ai-for-dummies/videos/browser-pt.jpg" width="1280" height="720" muted loop playsinline preload="none" aria-label="08 / Modelo no navegador. Um modelo de linguagem dentro da página. Os pesos baixam uma vez, do Hub. Na próxima visita, vêm do cache do navegador. Ele roda na sua própria GPU. Seu prompt nunca sai da página. O modelo vem até você." data-astro-cid-5bv36nto></video> <button class="toggle" type="button" data-film-toggle aria-pressed="false" data-astro-cid-5bv36nto> <svg class="icon-play" viewBox="0 0 16 16" aria-hidden="true" data-astro-cid-5bv36nto><path d="M4 2.5v11l9-5.5z" fill="currentColor" data-astro-cid-5bv36nto></path></svg> <svg class="icon-pause" viewBox="0 0 16 16" aria-hidden="true" data-astro-cid-5bv36nto><path d="M3.5 2.5h3v11h-3zm6 0h3v11h-3z" fill="currentColor" data-astro-cid-5bv36nto></path></svg> <span class="sr-only" data-language-content="en" data-astro-cid-5bv36nto> <span class="when-paused" data-astro-cid-5bv36nto>Play film</span> <span class="when-playing" data-astro-cid-5bv36nto>Pause film</span> </span><span class="sr-only" data-language-content="pt" hidden data-astro-cid-5bv36nto> <span class="when-paused" data-astro-cid-5bv36nto>Reproduzir filme</span> <span class="when-playing" data-astro-cid-5bv36nto>Pausar filme</span> </span> </button> <span class="progress" aria-hidden="true" data-astro-cid-5bv36nto><span data-film-progress data-astro-cid-5bv36nto></span></span> </div> <figcaption data-astro-cid-5bv36nto> <span class="meta" data-language-content="en" data-astro-cid-5bv36nto> Film · 24 s · silent, captioned </span><span class="meta" data-language-content="pt" hidden data-astro-cid-5bv36nto> Filme · 24 s · sem som, com legendas </span> <details data-astro-cid-5bv36nto> <summary data-astro-cid-5bv36nto> <span data-language-content="en" data-astro-cid-5bv36nto> Transcript </span><span data-language-content="pt" hidden data-astro-cid-5bv36nto> Transcrição </span> </summary> <ol data-language-content="en" data-astro-cid-5bv36nto> <li data-astro-cid-5bv36nto>A language model, inside a web page.</li><li data-astro-cid-5bv36nto>The weights download once, from the Hub.</li><li data-astro-cid-5bv36nto>Next visit, they load from the browser cache.</li><li data-astro-cid-5bv36nto>It runs on your own GPU.</li><li data-astro-cid-5bv36nto>Your prompt never leaves the page.</li><li data-astro-cid-5bv36nto>The model comes to you.</li> </ol><ol data-language-content="pt" hidden data-astro-cid-5bv36nto> <li data-astro-cid-5bv36nto>Um modelo de linguagem dentro da página.</li><li data-astro-cid-5bv36nto>Os pesos baixam uma vez, do Hub.</li><li data-astro-cid-5bv36nto>Na próxima visita, vêm do cache do navegador.</li><li data-astro-cid-5bv36nto>Ele roda na sua própria GPU.</li><li data-astro-cid-5bv36nto>Seu prompt nunca sai da página.</li><li data-astro-cid-5bv36nto>O modelo vem até você.</li> </ol> </details> </figcaption> </figure> <script type="module">const o=()=>window.matchMedia("(prefers-reduced-motion: reduce)").matches;function l(e){return e.videos.find(t=>!t.hidden)}function a(e){const t=e.videos.some(s=>!s.paused);e.root.dataset.paused=String(!t),e.button.setAttribute("aria-pressed",String(t))}function u(e){e.videos.forEach(s=>{s.hidden&&s.pause()});const t=l(e);t&&(t.preload="auto",t.play().catch(()=>a(e)))}function c(e){e.videos.forEach(t=>t.pause())}function d(e){e.inView&&!e.userPaused&&!o()?u(e):c(e)}function f(){const e=[];if(document.querySelectorAll("[data-film]").forEach(t=>{if(t.dataset.filmWired)return;t.dataset.filmWired="true";const s=t.querySelector("[data-film-toggle]");if(!s)return;const n={root:t,videos:[...t.querySelectorAll("video")],bar:t.querySelector("[data-film-progress]"),button:s,inView:!1,userPaused:o()};n.videos.forEach(r=>{r.addEventListener("play",()=>a(n)),r.addEventListener("pause",()=>a(n)),r.addEventListener("timeupdate",()=>{!n.bar||!r.duration||(n.bar.style.transform=`scaleX(${r.currentTime/r.duration})`)})}),s.addEventListener("click",()=>{const r=n.videos.some(i=>!i.paused);n.userPaused=r,r?c(n):u(n)}),e.push(n)}),!!e.length){if("IntersectionObserver"in window){const t=new IntersectionObserver(s=>{s.forEach(n=>{const r=e.find(i=>i.root===n.target);r&&(r.inView=n.isIntersecting,d(r))})},{threshold:.4});e.forEach(s=>t.observe(s.root))}window.addEventListener("ai-for-dummies:languagechange",()=>e.forEach(d))}}f();</script> </div> </section> <div class="model-lab" data-model-lab data-astro-cid-vutq5crx> <section class="lab-panel reveal" aria-labelledby="lab-chat-title" data-astro-cid-vutq5crx> <header class="lab-head" data-astro-cid-vutq5crx> <p class="eyebrow" data-language-content="en">01 · Chat</p> <p class="eyebrow" data-language-content="pt" hidden>01 · Conversa</p> <h2 id="lab-chat-title" data-astro-cid-vutq5crx> <span data-language-content="en" data-astro-cid-vutq5crx>Pick a model. <em data-astro-cid-vutq5crx>It downloads once.</em></span> <span data-language-content="pt" hidden data-astro-cid-vutq5crx>Escolha um modelo. <em data-astro-cid-vutq5crx>Ele baixa uma vez.</em></span> </h2> <p class="lab-lede" data-language-content="en">These are small open models, a few hundred megabytes each, so expect short answers and real mistakes. That is the point: you can feel what size buys. The size shown is what your browser will download; after that the model loads from its cache.</p> <p class="lab-lede" data-language-content="pt" hidden>São modelos abertos pequenos, de algumas centenas de megabytes, então espere respostas curtas e erros de verdade. É esse o ponto: dá para sentir o que o tamanho compra. O tamanho mostrado é o que seu navegador vai baixar; depois disso o modelo carrega do cache.</p> </header> <div class="lab-body" data-astro-cid-vutq5crx> <fieldset class="model-list" data-lab-models data-astro-cid-vutq5crx> <legend data-astro-cid-vutq5crx> <span data-language-content="en">Model</span> <span data-language-content="pt" hidden>Modelo</span> </legend> <label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="HuggingFaceTB/SmolLM2-360M-Instruct" checked data-webgpu-dtype="q4f16" data-webgpu-mb="276" data-wasm-dtype="quantized" data-wasm-mb="368" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>SmolLM2 360M Instruct</span> <span class="option-meta" data-astro-cid-vutq5crx> Hugging Face · 360M · <span data-size data-astro-cid-vutq5crx>276 MB</span> </span> <span class="option-note" data-language-content="en">The default. Coherent short answers in English; quick on most laptops.</span> <span class="option-note" data-language-content="pt" hidden>O padrão. Respostas curtas e coerentes em inglês; rápido na maioria dos notebooks.</span> </span> </label><label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="HuggingFaceTB/SmolLM2-135M-Instruct" data-webgpu-dtype="q4f16" data-webgpu-mb="121" data-wasm-dtype="quantized" data-wasm-mb="140" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>SmolLM2 135M Instruct</span> <span class="option-meta" data-astro-cid-vutq5crx> Hugging Face · 135M · <span data-size data-astro-cid-vutq5crx>121 MB</span> </span> <span class="option-note" data-language-content="en">The smallest. Fast even on a CPU, and visibly wrong often: a good lesson in itself.</span> <span class="option-note" data-language-content="pt" hidden>O menor. Rápido até na CPU e erra com frequência: uma boa lição por si só.</span> </span> </label><label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="onnx-community/gemma-3-270m-it-ONNX" data-webgpu-dtype="q4f16" data-webgpu-mb="293" data-wasm-dtype="q4" data-wasm-mb="344" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>Gemma 3 270M IT</span> <span class="option-meta" data-astro-cid-vutq5crx> Google · 270M · <span data-size data-astro-cid-vutq5crx>293 MB</span> </span> <span class="option-note" data-language-content="en">Google's smallest open model. Multilingual, so it can answer in Portuguese.</span> <span class="option-note" data-language-content="pt" hidden>O menor modelo aberto do Google. Multilíngue, então responde em português.</span> </span> </label><label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="onnx-community/LFM2-350M-ONNX" data-webgpu-dtype="q4f16" data-webgpu-mb="259" data-wasm-dtype="q4" data-wasm-mb="297" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>LFM2 350M</span> <span class="option-meta" data-astro-cid-vutq5crx> Liquid AI · 350M · <span data-size data-astro-cid-vutq5crx>259 MB</span> </span> <span class="option-note" data-language-content="en">A hybrid (not pure transformer) design built for on-device speed.</span> <span class="option-note" data-language-content="pt" hidden>Arquitetura híbrida (não só transformer) feita para velocidade no dispositivo.</span> </span> </label><label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="onnx-community/Qwen2.5-0.5B-Instruct" data-webgpu-dtype="q4f16" data-webgpu-mb="493" data-wasm-dtype="quantized" data-wasm-mb="522" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>Qwen2.5 0.5B Instruct</span> <span class="option-meta" data-astro-cid-vutq5crx> Alibaba Qwen · 0.5B · <span data-size data-astro-cid-vutq5crx>493 MB</span> </span> <span class="option-note" data-language-content="en">Noticeably better at instructions and code; a bigger download.</span> <span class="option-note" data-language-content="pt" hidden>Bem melhor em instruções e código; download maior.</span> </span> </label><label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="onnx-community/Qwen3-0.6B-ONNX" data-webgpu-dtype="q4f16" data-webgpu-mb="582" data-wasm-dtype="quantized" data-wasm-mb="630" data-thinking="true" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>Qwen3 0.6B</span> <span class="option-meta" data-astro-cid-vutq5crx> Alibaba Qwen · 0.6B · <span data-size data-astro-cid-vutq5crx>582 MB</span> </span> <span class="option-note" data-language-content="en">Can think before answering. Turn thinking on to watch effort cost time.</span> <span class="option-note" data-language-content="pt" hidden>Pode pensar antes de responder. Ligue o raciocínio para ver o esforço custar tempo.</span> </span> </label> <label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="custom" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx> <span data-language-content="en">Any model from the Hub</span> <span data-language-content="pt" hidden>Qualquer modelo do Hub</span> </span> <span class="option-note" data-language-content="en">An ONNX text-generation repo, such as one from onnx-community. The page checks its size first.</span> <span class="option-note" data-language-content="pt" hidden>Um repositório ONNX de geração de texto, como os da onnx-community. A página confere o tamanho antes.</span> <input class="custom-id" type="text" data-lab-custom placeholder="onnx-community/…" spellcheck="false" autocomplete="off" aria-label="Hugging Face model id" data-astro-cid-vutq5crx> </span> </label> <label class="model-option" data-builtin-option hidden data-astro-cid-vutq5crx> <input type="radio" name="lab-model" value="builtin" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx> <span data-language-content="en">Built into this browser</span> <span data-language-content="pt" hidden>Embutido neste navegador</span> </span> <span class="option-note" data-language-content="en">Your browser ships its own on-device model (Chrome's Prompt API). The browser downloads and manages it, not this page.</span> <span class="option-note" data-language-content="pt" hidden>Seu navegador traz o próprio modelo local (Prompt API do Chrome). Quem baixa e gerencia é o navegador, não esta página.</span> </span> </label> </fieldset> <div class="console" data-astro-cid-vutq5crx> <p class="device" data-lab-device data-astro-cid-vutq5crx>…</p> <div class="load-row" data-astro-cid-vutq5crx> <button class="primary" type="button" data-lab-load data-astro-cid-vutq5crx> <span class="spin" data-astro-cid-vutq5crx><svg viewBox="0 0 16 16" aria-hidden="true"><path d="M2 8a6 6 0 0 1 6-6" fill="none" stroke="currentColor" stroke-width="2"/></svg></span> <span data-lab-load-label data-astro-cid-vutq5crx>…</span> </button> <label class="check" data-astro-cid-vutq5crx> <input type="checkbox" data-lab-cpu data-astro-cid-vutq5crx> <span data-language-content="en">Force CPU</span> <span data-language-content="pt" hidden>Forçar CPU</span> </label> </div> <div class="bar" role="progressbar" aria-valuemin="0" aria-valuemax="100" aria-valuenow="0" aria-label="Download progress" data-lab-bar data-astro-cid-vutq5crx> <span data-astro-cid-vutq5crx></span> </div> <p class="status" data-lab-status aria-live="polite" data-astro-cid-vutq5crx></p> <ol class="chat-log" data-lab-log aria-label="Conversation" data-astro-cid-vutq5crx></ol> <form class="prompt" data-lab-form data-astro-cid-vutq5crx> <label class="sr-only" for="lab-prompt" data-astro-cid-vutq5crx>Prompt</label> <textarea id="lab-prompt" rows="3" disabled data-lab-prompt data-astro-cid-vutq5crx></textarea> <div class="samples" data-astro-cid-vutq5crx> <button type="button" class="chip" data-lab-sample disabled data-en="Explain a git worktree in two sentences." data-pt="Explique uma git worktree em duas frases." data-astro-cid-vutq5crx> <span data-language-content="en">Explain a git worktree in two sentences.</span> <span data-language-content="pt" hidden>Explique uma git worktree em duas frases.</span> </button><button type="button" class="chip" data-lab-sample disabled data-en="Write a haiku about code review." data-pt="Escreva um haicai sobre code review." data-astro-cid-vutq5crx> <span data-language-content="en">Write a haiku about code review.</span> <span data-language-content="pt" hidden>Escreva um haicai sobre code review.</span> </button><button type="button" class="chip" data-lab-sample disabled data-en="What is 17 × 23? Show your steps." data-pt="Quanto é 17 × 23? Mostre os passos." data-astro-cid-vutq5crx> <span data-language-content="en">What is 17 × 23? Show your steps.</span> <span data-language-content="pt" hidden>Quanto é 17 × 23? Mostre os passos.</span> </button> </div> <div class="prompt-actions" data-astro-cid-vutq5crx> <label class="check" data-lab-thinking-wrap hidden data-astro-cid-vutq5crx> <input type="checkbox" data-lab-thinking data-astro-cid-vutq5crx> <span data-language-content="en">Think first</span> <span data-language-content="pt" hidden>Pensar antes</span> </label> <button type="button" class="ghost" data-lab-reset disabled data-astro-cid-vutq5crx> <span data-language-content="en">New chat</span> <span data-language-content="pt" hidden>Nova conversa</span> </button> <button type="button" class="ghost" data-lab-stop disabled data-astro-cid-vutq5crx> <span data-language-content="en">Stop</span> <span data-language-content="pt" hidden>Parar</span> </button> <button type="submit" class="primary" data-lab-send disabled data-astro-cid-vutq5crx> <span data-language-content="en">Send</span> <span data-language-content="pt" hidden>Enviar</span> </button> </div> </form> <p class="stats" data-lab-stats aria-live="polite" data-astro-cid-vutq5crx></p> </div> </div> </section> <section class="lab-panel reveal" aria-labelledby="lab-search-title" data-embed-lab data-astro-cid-vutq5crx> <header class="lab-head" data-astro-cid-vutq5crx> <p class="eyebrow" data-language-content="en">02 · Search by meaning</p> <p class="eyebrow" data-language-content="pt" hidden>02 · Busca por significado</p> <h2 id="lab-search-title" data-astro-cid-vutq5crx> <span data-language-content="en" data-astro-cid-vutq5crx>Words differ. <em data-astro-cid-vutq5crx>Meaning matches.</em></span> <span data-language-content="pt" hidden data-astro-cid-vutq5crx>As palavras mudam. <em data-astro-cid-vutq5crx>O sentido combina.</em></span> </h2> <p class="lab-lede" data-language-content="en">This is the retrieval half of RAG. An embedding model turns each sentence into a list of numbers; sentences that mean similar things land close together, even with no words in common. Ask about "vacation days" and the sentence about "paid leave" should rise to the top.</p> <p class="lab-lede" data-language-content="pt" hidden>Esta é a metade de recuperação do RAG. Um modelo de embeddings transforma cada frase em uma lista de números; frases com sentido parecido ficam perto, mesmo sem palavras em comum. Pergunte sobre "dias de folga" e a frase sobre "férias" deve subir para o topo.</p> </header> <div class="lab-body" data-astro-cid-vutq5crx> <fieldset class="model-list compact" data-astro-cid-vutq5crx> <legend data-astro-cid-vutq5crx> <span data-language-content="en">Embedding model</span> <span data-language-content="pt" hidden>Modelo de embeddings</span> </legend> <label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-embed" value="Xenova/all-MiniLM-L6-v2" checked data-dtype="quantized" data-mb="24" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>all-MiniLM-L6-v2</span> <span class="option-meta" data-astro-cid-vutq5crx> <span data-language-content="en">English</span> <span data-language-content="pt" hidden>Inglês</span> · 24 MB
|
||
</span> </span> </label><label class="model-option" data-astro-cid-vutq5crx> <input type="radio" name="lab-embed" value="Xenova/paraphrase-multilingual-MiniLM-L12-v2" data-dtype="quantized" data-mb="150" data-multilingual="true" data-astro-cid-vutq5crx> <span class="option-main" data-astro-cid-vutq5crx> <span class="option-name" data-astro-cid-vutq5crx>paraphrase-multilingual-MiniLM-L12-v2</span> <span class="option-meta" data-astro-cid-vutq5crx> <span data-language-content="en">50+ languages, Portuguese included</span> <span data-language-content="pt" hidden>50+ idiomas, incluindo português</span> · 150 MB
|
||
</span> </span> </label> <div class="load-row" data-astro-cid-vutq5crx> <button class="primary" type="button" data-embed-load data-astro-cid-vutq5crx> <span class="spin" data-astro-cid-vutq5crx><svg viewBox="0 0 16 16" aria-hidden="true"><path d="M2 8a6 6 0 0 1 6-6" fill="none" stroke="currentColor" stroke-width="2"/></svg></span> <span data-embed-load-label data-astro-cid-vutq5crx>…</span> </button> </div> <div class="bar" role="progressbar" aria-valuemin="0" aria-valuemax="100" aria-valuenow="0" aria-label="Download progress" data-embed-bar data-astro-cid-vutq5crx> <span data-astro-cid-vutq5crx></span> </div> <p class="status" data-embed-status aria-live="polite" data-astro-cid-vutq5crx></p> </fieldset> <div class="console" data-astro-cid-vutq5crx> <form class="query" data-embed-form data-astro-cid-vutq5crx> <label class="sr-only" for="lab-query" data-astro-cid-vutq5crx>Query</label> <input id="lab-query" type="search" disabled data-embed-query autocomplete="off" data-astro-cid-vutq5crx> <button type="submit" class="primary" disabled data-embed-go data-astro-cid-vutq5crx> <span data-language-content="en">Rank</span> <span data-language-content="pt" hidden>Ordenar</span> </button> </form> <div class="samples" data-astro-cid-vutq5crx> <button type="button" class="chip" data-embed-sample disabled data-en="How many vacation days do I have?" data-pt="Quantos dias de folga eu tenho?" data-astro-cid-vutq5crx> <span data-language-content="en">How many vacation days do I have?</span> <span data-language-content="pt" hidden>Quantos dias de folga eu tenho?</span> </button><button type="button" class="chip" data-embed-sample disabled data-en="Two agents keep overwriting each other" data-pt="Dois agentes ficam sobrescrevendo um ao outro" data-astro-cid-vutq5crx> <span data-language-content="en">Two agents keep overwriting each other</span> <span data-language-content="pt" hidden>Dois agentes ficam sobrescrevendo um ao outro</span> </button><button type="button" class="chip" data-embed-sample disabled data-en="Stop the model from making things up" data-pt="Impedir o modelo de inventar coisas" data-astro-cid-vutq5crx> <span data-language-content="en">Stop the model from making things up</span> <span data-language-content="pt" hidden>Impedir o modelo de inventar coisas</span> </button> </div> <ol class="ranked" data-embed-list aria-live="polite" data-astro-cid-vutq5crx> <li data-index="0" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">Use a fast model for small, well-specified edits.</span> <span data-language-content="pt" hidden>Use um modelo rápido para edições pequenas e bem especificadas.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li><li data-index="1" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">Raise reasoning effort when the task is ambiguous or costly to get wrong.</span> <span data-language-content="pt" hidden>Aumente o esforço de raciocínio quando a tarefa for ambígua ou cara de errar.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li><li data-index="2" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">Give each agent its own git worktree so edits never collide.</span> <span data-language-content="pt" hidden>Dê a cada agente sua própria git worktree para as edições nunca colidirem.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li><li data-index="3" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">A skill keeps only its description in context until a request matches it.</span> <span data-language-content="pt" hidden>Uma skill mantém só a descrição no contexto até um pedido combinar com ela.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li><li data-index="4" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">A pre-push hook runs the checks before code leaves your machine.</span> <span data-language-content="pt" hidden>Um hook de pre-push roda as checagens antes do código sair da sua máquina.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li><li data-index="5" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">Split documents into labelled chunks and cite the source of every answer.</span> <span data-language-content="pt" hidden>Divida documentos em trechos rotulados e cite a fonte de cada resposta.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li><li data-index="6" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">If no chunk scores high enough, answer that the sources do not say.</span> <span data-language-content="pt" hidden>Se nenhum trecho pontuar o suficiente, responda que as fontes não dizem.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li><li data-index="7" data-astro-cid-vutq5crx> <span class="score" data-score data-astro-cid-vutq5crx>
|
||
–
|
||
</span> <span class="line" data-astro-cid-vutq5crx> <span data-language-content="en">Full-time staff get 25 working days of paid leave a year.</span> <span data-language-content="pt" hidden>Quem trabalha em tempo integral tem 25 dias úteis de férias por ano.</span> <span class="meter" aria-hidden="true" data-astro-cid-vutq5crx> <span data-meter data-astro-cid-vutq5crx></span> </span> </span> </li> </ol> <script type="application/json" data-embed-corpus>[{"en":"Use a fast model for small, well-specified edits.","pt":"Use um modelo rápido para edições pequenas e bem especificadas."},{"en":"Raise reasoning effort when the task is ambiguous or costly to get wrong.","pt":"Aumente o esforço de raciocínio quando a tarefa for ambígua ou cara de errar."},{"en":"Give each agent its own git worktree so edits never collide.","pt":"Dê a cada agente sua própria git worktree para as edições nunca colidirem."},{"en":"A skill keeps only its description in context until a request matches it.","pt":"Uma skill mantém só a descrição no contexto até um pedido combinar com ela."},{"en":"A pre-push hook runs the checks before code leaves your machine.","pt":"Um hook de pre-push roda as checagens antes do código sair da sua máquina."},{"en":"Split documents into labelled chunks and cite the source of every answer.","pt":"Divida documentos em trechos rotulados e cite a fonte de cada resposta."},{"en":"If no chunk scores high enough, answer that the sources do not say.","pt":"Se nenhum trecho pontuar o suficiente, responda que as fontes não dizem."},{"en":"Full-time staff get 25 working days of paid leave a year.","pt":"Quem trabalha em tempo integral tem 25 dias úteis de férias por ano."}]</script> </div> </div> </section> <section class="lab-panel reveal" aria-labelledby="lab-disk-title" data-astro-cid-vutq5crx> <header class="lab-head" data-astro-cid-vutq5crx> <p class="eyebrow" data-language-content="en">03 · Your disk</p> <p class="eyebrow" data-language-content="pt" hidden>03 · Seu disco</p> <h2 id="lab-disk-title" data-astro-cid-vutq5crx> <span data-language-content="en" data-astro-cid-vutq5crx>What stays <em data-astro-cid-vutq5crx>behind.</em></span> <span data-language-content="pt" hidden data-astro-cid-vutq5crx>O que <em data-astro-cid-vutq5crx>fica.</em></span> </h2> </header> <div class="lab-body disk" data-astro-cid-vutq5crx> <p class="lab-lede" data-language-content="en">Downloaded weights live in this site's browser storage, not in a folder you can see. They stay until you clear them here or clear site data. Nothing you type is stored or sent anywhere.</p> <p class="lab-lede" data-language-content="pt" hidden>Os pesos baixados ficam no armazenamento do navegador para este site, não numa pasta visível. Ficam até você limpar aqui ou apagar os dados do site. Nada do que você digita é guardado ou enviado.</p> <div class="disk-row" data-astro-cid-vutq5crx> <p class="status" data-disk-usage aria-live="polite" data-astro-cid-vutq5crx>…</p> <button type="button" class="ghost" data-disk-clear data-astro-cid-vutq5crx> <span data-language-content="en">Delete downloaded models</span> <span data-language-content="pt" hidden>Apagar modelos baixados</span> </button> </div> </div> </section> </div> <script type="module" src="/ai-for-dummies/_astro/ModelLab.astro_astro_type_script_index_0_lang.-kZIlw8W.js"></script> <section class="pipeline reveal" data-astro-cid-t3nv4xef> <div class="depth-slow" data-astro-cid-t3nv4xef> <p class="eyebrow" data-language-content="en">What happens on Load</p> <p class="eyebrow" data-language-content="pt" hidden>O que acontece ao carregar</p> <h2 data-language-content="en" data-astro-cid-t3nv4xef>Four steps,<br><em>all in your browser.</em></h2> <h2 data-language-content="pt" hidden data-astro-cid-t3nv4xef>Quatro passos,<br><em>todos no navegador.</em></h2> <p data-language-content="en" data-astro-cid-t3nv4xef>The page itself is static. The work happens in a background worker, so the page stays responsive while the model thinks.</p> <p data-language-content="pt" hidden data-astro-cid-t3nv4xef>A página em si é estática. O trabalho acontece num worker em segundo plano, então a página continua responsiva enquanto o modelo pensa.</p> <svg class="draw-line" viewBox="0 0 4 110" aria-hidden="true" focusable="false" style="--draw-height: 110px" data-astro-cid-n6ozym62> <path d="M 0.5 0 V 110" pathLength="100" data-astro-cid-n6ozym62></path> </svg> </div> <ol class="flow depth-fast" data-astro-cid-ld3fmquo> <li data-astro-cid-ld3fmquo> <article data-astro-cid-ld3fmquo> <b data-astro-cid-ld3fmquo>01</b> <div data-astro-cid-ld3fmquo> <strong data-language-content="en" data-astro-cid-ld3fmquo>Detect</strong> <strong data-language-content="pt" hidden data-astro-cid-ld3fmquo>Detectar</strong> <span data-language-content="en" data-astro-cid-ld3fmquo>Check for WebGPU with half-precision shaders. Without it, fall back to WebAssembly on the CPU.</span> <span data-language-content="pt" hidden data-astro-cid-ld3fmquo>Confere se há WebGPU com shaders de meia precisão. Sem isso, volta para WebAssembly na CPU.</span> </div> </article> </li><li data-astro-cid-ld3fmquo> <article data-astro-cid-ld3fmquo> <b data-astro-cid-ld3fmquo>02</b> <div data-astro-cid-ld3fmquo> <strong data-language-content="en" data-astro-cid-ld3fmquo>Download</strong> <strong data-language-content="pt" hidden data-astro-cid-ld3fmquo>Baixar</strong> <span data-language-content="en" data-astro-cid-ld3fmquo>Fetch the quantized ONNX weights and tokenizer from the Hub, once, into the browser cache.</span> <span data-language-content="pt" hidden data-astro-cid-ld3fmquo>Busca os pesos ONNX quantizados e o tokenizador no Hub, uma vez, para o cache do navegador.</span> </div> </article> </li><li data-astro-cid-ld3fmquo> <article data-astro-cid-ld3fmquo> <b data-astro-cid-ld3fmquo>03</b> <div data-astro-cid-ld3fmquo> <strong data-language-content="en" data-astro-cid-ld3fmquo>Compile</strong> <strong data-language-content="pt" hidden data-astro-cid-ld3fmquo>Compilar</strong> <span data-language-content="en" data-astro-cid-ld3fmquo>ONNX Runtime Web builds the model's graph for your GPU or CPU. This is the pause after the bar fills.</span> <span data-language-content="pt" hidden data-astro-cid-ld3fmquo>O ONNX Runtime Web monta o grafo do modelo para sua GPU ou CPU. É a pausa depois que a barra enche.</span> </div> </article> </li><li data-astro-cid-ld3fmquo> <article data-astro-cid-ld3fmquo> <b data-astro-cid-ld3fmquo>04</b> <div data-astro-cid-ld3fmquo> <strong data-language-content="en" data-astro-cid-ld3fmquo>Stream</strong> <strong data-language-content="pt" hidden data-astro-cid-ld3fmquo>Transmitir</strong> <span data-language-content="en" data-astro-cid-ld3fmquo>Each new token is sent back to the page as it is produced, so you watch the answer being written.</span> <span data-language-content="pt" hidden data-astro-cid-ld3fmquo>Cada novo token volta para a página assim que é gerado, então você vê a resposta sendo escrita.</span> </div> </article> </li> </ol> </section> <section class="sources reveal" aria-labelledby="lab-sources-title" data-astro-cid-t3nv4xef> <p class="eyebrow" data-language-content="en">Other places</p> <p class="eyebrow" data-language-content="pt" hidden>Outros lugares</p> <h2 id="lab-sources-title" data-astro-cid-t3nv4xef> <span data-language-content="en" data-astro-cid-t3nv4xef>Where open models <em data-astro-cid-t3nv4xef>come from.</em></span> <span data-language-content="pt" hidden data-astro-cid-t3nv4xef>De onde vêm os <em data-astro-cid-t3nv4xef>modelos abertos.</em></span> </h2> <div class="grid" data-astro-cid-t3nv4xef> <article class="card" data-astro-cid-t3nv4xef> <strong class="card-label" data-language-content="en">Hub</strong> <strong class="card-label" data-language-content="pt" hidden>Hub</strong> <h3 data-language-content="en">Hugging Face</h3> <h3 data-language-content="pt" hidden>Hugging Face</h3> <p data-language-content="en">Where this page gets its weights. Look for repos with an onnx/ folder, often under onnx-community, to run them in a browser.</p> <p data-language-content="pt" hidden>De onde esta página tira os pesos. Procure repositórios com pasta onnx/, muitas vezes na onnx-community, para rodar no navegador.</p> <a href="https://huggingface.co/models?library=transformers.js&sort=trending" target="_blank" rel="noreferrer" data-astro-cid-t3nv4xef> <span data-language-content="en">Browse browser-ready models ↗</span> <span data-language-content="pt" hidden>Ver modelos prontos para navegador ↗</span> </a> </article><article class="card" data-astro-cid-t3nv4xef> <strong class="card-label" data-language-content="en">Browser</strong> <strong class="card-label" data-language-content="pt" hidden>Navegador</strong> <h3 data-language-content="en">Built-in models</h3> <h3 data-language-content="pt" hidden>Modelos embutidos</h3> <p data-language-content="en">Some browsers ship their own on-device model behind a Prompt API. When yours does, the lab offers it as an extra option.</p> <p data-language-content="pt" hidden>Alguns navegadores trazem o próprio modelo local via Prompt API. Quando o seu traz, o laboratório oferece como opção extra.</p> <a href="https://developer.chrome.com/docs/ai/prompt-api" target="_blank" rel="noreferrer" data-astro-cid-t3nv4xef> <span data-language-content="en">Read the Prompt API ↗</span> <span data-language-content="pt" hidden>Ler sobre a Prompt API ↗</span> </a> </article><article class="card" data-astro-cid-t3nv4xef> <strong class="card-label" data-language-content="en">Local</strong> <strong class="card-label" data-language-content="pt" hidden>Local</strong> <h3 data-language-content="en">Your own machine</h3> <h3 data-language-content="pt" hidden>Sua própria máquina</h3> <p data-language-content="en">For models too big for a tab, run them as a local server with Ollama or llama.cpp; the same open weights, more memory.</p> <p data-language-content="pt" hidden>Para modelos grandes demais para uma aba, rode como servidor local com Ollama ou llama.cpp; os mesmos pesos abertos, mais memória.</p> <a href="https://ollama.com/library" target="_blank" rel="noreferrer" data-astro-cid-t3nv4xef> <span data-language-content="en">Ollama library ↗</span> <span data-language-content="pt" hidden>Biblioteca do Ollama ↗</span> </a> </article> </div> </section> </main> <section class="footer" data-astro-cid-bmvnf73n> <nav class="links" aria-label="Chapter navigation" data-astro-cid-bmvnf73n> <nav class="links" aria-label="Chapter navigation" data-astro-cid-t3nv4xef> <a href="/ai-for-dummies/rag/" data-astro-cid-t3nv4xef>← RAG</a> <a href="/ai-for-dummies/films/" data-astro-cid-t3nv4xef>Films →</a> </nav> </nav> <div class="text" data-astro-cid-bmvnf73n> </div> </section> </body></html> |