Add Ollama support, progress bar, and professional UI redesign
This commit is contained in:
119
sample_files/webllm.html
Normal file
119
sample_files/webllm.html
Normal file
@@ -0,0 +1,119 @@
|
||||
<!DOCTYPE html>
|
||||
<html lang="fr">
|
||||
<head>
|
||||
<meta charset="UTF-8">
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
||||
<title>Test LLM Local - WebGPU</title>
|
||||
<style>
|
||||
body { font-family: system-ui, sans-serif; max-width: 800px; margin: 2rem auto; padding: 0 1rem; line-height: 1.5; }
|
||||
#chat-box { border: 1px solid #ccc; padding: 1rem; height: 400px; overflow-y: auto; border-radius: 8px; background: #f9f9f9; margin-bottom: 1rem; }
|
||||
.message { margin-bottom: 1rem; padding: 0.5rem 1rem; border-radius: 8px; }
|
||||
.user { background: #e3f2fd; align-self: flex-end; text-align: right; }
|
||||
.bot { background: #fff; border: 1px solid #eee; }
|
||||
#controls { display: flex; gap: 10px; }
|
||||
input { flex-grow: 1; padding: 10px; border-radius: 4px; border: 1px solid #ddd; }
|
||||
button { padding: 10px 20px; background: #007bff; color: white; border: none; border-radius: 4px; cursor: pointer; }
|
||||
button:disabled { background: #ccc; }
|
||||
#status { font-size: 0.9rem; color: #666; margin-bottom: 1rem; font-style: italic; }
|
||||
</style>
|
||||
</head>
|
||||
<body>
|
||||
|
||||
<h2>🤖 Mon LLM Local (via Chrome WebGPU)</h2>
|
||||
|
||||
<div id="status">Initialisation... cliquez sur "Charger le modèle" pour commencer.</div>
|
||||
|
||||
<div id="chat-box"></div>
|
||||
|
||||
<div id="controls">
|
||||
<input type="text" id="user-input" placeholder="Écrivez votre message ici..." disabled>
|
||||
<button id="send-btn" disabled>Envoyer</button>
|
||||
</div>
|
||||
|
||||
<button id="load-btn" style="margin-top: 10px; background-color: #28a745;">Charger le Modèle (Llama 3.2)</button>
|
||||
|
||||
<script type="module">
|
||||
// Importation de WebLLM directement depuis le CDN
|
||||
import { CreateMLCEngine } from "https://esm.run/@mlc-ai/web-llm";
|
||||
|
||||
// Configuration du modèle (ici Llama 3.2 1B, léger et rapide)
|
||||
const selectedModel = "Llama-3.2-1B-Instruct-q4f16_1-MLC";
|
||||
|
||||
let engine;
|
||||
const statusLabel = document.getElementById('status');
|
||||
const chatBox = document.getElementById('chat-box');
|
||||
const userInput = document.getElementById('user-input');
|
||||
const sendBtn = document.getElementById('send-btn');
|
||||
const loadBtn = document.getElementById('load-btn');
|
||||
|
||||
// Fonction pour mettre à jour l'état de chargement
|
||||
const initProgressCallback = (report) => {
|
||||
statusLabel.innerText = report.text;
|
||||
};
|
||||
|
||||
// 1. Chargement du moteur (Engine)
|
||||
loadBtn.addEventListener('click', async () => {
|
||||
loadBtn.disabled = true;
|
||||
statusLabel.innerText = "Démarrage du téléchargement du modèle (peut prendre quelques minutes)...";
|
||||
|
||||
try {
|
||||
// Création du moteur WebLLM
|
||||
engine = await CreateMLCEngine(
|
||||
selectedModel,
|
||||
{ initProgressCallback: initProgressCallback }
|
||||
);
|
||||
|
||||
statusLabel.innerText = "✅ Modèle chargé et prêt ! (GPU Actif)";
|
||||
userInput.disabled = false;
|
||||
sendBtn.disabled = false;
|
||||
loadBtn.style.display = 'none';
|
||||
userInput.focus();
|
||||
} catch (err) {
|
||||
statusLabel.innerText = "❌ Erreur : " + err.message;
|
||||
loadBtn.disabled = false;
|
||||
}
|
||||
});
|
||||
|
||||
// 2. Fonction d'envoi de message
|
||||
sendBtn.addEventListener('click', sendMessage);
|
||||
userInput.addEventListener('keypress', (e) => { if(e.key === 'Enter') sendMessage() });
|
||||
|
||||
async function sendMessage() {
|
||||
const text = userInput.value.trim();
|
||||
if (!text) return;
|
||||
|
||||
// Affichage message utilisateur
|
||||
appendMessage(text, 'user');
|
||||
userInput.value = '';
|
||||
|
||||
// Création placeholder pour la réponse
|
||||
const botMessageDiv = appendMessage("...", 'bot');
|
||||
let fullResponse = "";
|
||||
|
||||
// Inférence (Génération)
|
||||
const chunks = await engine.chat.completions.create({
|
||||
messages: [{ role: "user", content: text }],
|
||||
stream: true, // Important pour voir le texte s'écrire en temps réel
|
||||
});
|
||||
|
||||
// Lecture du flux (Streaming)
|
||||
for await (const chunk of chunks) {
|
||||
const content = chunk.choices[0]?.delta?.content || "";
|
||||
fullResponse += content;
|
||||
botMessageDiv.innerText = fullResponse;
|
||||
// Auto-scroll vers le bas
|
||||
chatBox.scrollTop = chatBox.scrollHeight;
|
||||
}
|
||||
}
|
||||
|
||||
function appendMessage(text, sender) {
|
||||
const div = document.createElement('div');
|
||||
div.classList.add('message', sender);
|
||||
div.innerText = text;
|
||||
chatBox.appendChild(div);
|
||||
chatBox.scrollTop = chatBox.scrollHeight;
|
||||
return div;
|
||||
}
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
Reference in New Issue
Block a user