Instructions to use Meanblock/JEV-CPU with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Meanblock/JEV-CPU with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("zero-shot-classification", model="Meanblock/JEV-CPU")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Meanblock/JEV-CPU", device_map="auto") - Notebooks
- Google Colab
- Kaggle
File size: 13,653 Bytes
7845694 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 | import { createApp, reactive } from "https://cdn.jsdelivr.net/npm/vue@3.5.21/dist/vue.esm-browser.prod.js";
const worker = new Worker("worker.js", { type: "module" });
const $ = (selector) => document.querySelector(selector);
const loadButton = $("#load");
const runButton = $("#run");
const modelSelect = $("#model-select");
const uiMode = $("#ui-mode");
const models = {
"qwen3-0.6b": {
name: "Qwen3 0.6B", short: "Qwen3 · 0.6B", size: "639 MB",
url: "https://huggingface.co/Qwen/Qwen3-0.6B-GGUF",
notice: "Smaller model optimized for small devices. Accuracy may be worse.", noticeClass: "mobile",
},
"minicpm5-2b": {
name: "MiniCPM5 2B", short: "MiniCPM5 · 2B", size: "1.56 GB",
url: "https://huggingface.co/openbmb/MiniCPM5-2B-GGUF",
notice: "Larger model. Loading may be slower or may not fit on some low-end devices.", noticeClass: "",
},
"qwen3.5-4b": {
name: "Qwen3.5 4B", short: "Qwen3.5 · 4B", size: "3.01 GB",
url: "https://huggingface.co/bartowski/Qwen_Qwen3.5-4B-GGUF",
notice: "High-memory desktop model. Allow several gigabytes of free GPU memory and browser storage.", noticeClass: "desktop-heavy",
},
};
const presets = {
account: {
state: "A customer says a password reset succeeded, but every login attempt still returns ‘account locked’. Two unlock emails were requested and neither arrived.",
question: "Which queue should handle this request?",
options: ["Account access support", "Billing support", "Close as resolved"],
},
email: {
state: "An email claims to be from the payroll team and says the recipient’s salary payment will be suspended today. It comes from payroll-review@outlook.com and links to a non-company sign-in page asking for a password and verification code.",
question: "How should this email be classified?",
options: ["Legitimate", "Spam", "Phishing"],
},
};
const MIN_OPTIONS = 2;
const MAX_OPTIONS = 20;
const optionList = $("#option-list");
const addOptionButton = $("#add-option");
const removeOptionButton = $("#remove-option");
const files = new Map();
let ready = false;
const isMobileDevice = navigator.userAgentData?.mobile === true
|| /Android|iPhone|iPad|iPod|Mobile/i.test(navigator.userAgent)
|| window.matchMedia("(max-width: 600px)").matches;
const supportState = reactive({ text: "Checking WebGPU…", kind: "", icon: "memory" });
createApp({ setup: () => supportState }).mount("#support");
function applyUiMode(plain) {
document.body.classList.toggle("plain-ui", plain);
uiMode.checked = plain;
try {
localStorage.setItem("semif-ui-mode", plain ? "plain" : "original");
} catch (_) {
// The preference is optional; inference does not depend on browser storage.
}
}
let savedUiMode = false;
try {
savedUiMode = localStorage.getItem("semif-ui-mode") === "plain";
} catch (_) {
// Some embedded browsers disable local storage.
}
applyUiMode(savedUiMode);
uiMode.addEventListener("change", () => applyUiMode(uiMode.checked));
function seconds(ms) {
return `${(ms / 1000).toFixed(3)} s`;
}
function setSupport(text, kind = "") {
supportState.text = text;
supportState.kind = kind;
supportState.icon = kind === "error" ? "error" : kind === "ok" ? "check_circle" : "memory";
}
function renderSelectedModel() {
const selected = models[modelSelect.value];
$("#selected-model").textContent = selected.short;
$("#model-size").textContent = `${selected.size} model`;
$("#model-link").href = selected.url;
$("#download-detail").textContent = `${selected.size} on first load`;
const notice = $("#model-notice");
notice.textContent = selected.notice;
notice.className = `model-notice ${selected.noticeClass}`.trim();
document.querySelectorAll("[data-quality-model]").forEach((row) => {
row.classList.toggle("selected", row.dataset.qualityModel === modelSelect.value);
});
loadButton.innerHTML = `<span class="material-symbols-rounded" aria-hidden="true">download</span> load ${selected.name}`;
}
function renderProgress(event) {
if (!event.file) return;
if (event.status === "progress" && Number.isFinite(event.loaded) && Number.isFinite(event.total)) {
files.set(event.file, { loaded: event.loaded, total: event.total });
} else if (event.status === "done" && files.has(event.file)) {
const item = files.get(event.file);
files.set(event.file, { loaded: item.total, total: item.total });
}
const totals = [...files.values()].reduce((sum, item) => ({ loaded: sum.loaded + item.loaded, total: sum.total + item.total }), { loaded: 0, total: 0 });
if (totals.total > 0) {
const percent = Math.min(100, (totals.loaded / totals.total) * 100);
$("#download-meter").style.width = `${percent}%`;
$("#download-value").textContent = `${percent.toFixed(0)}%`;
$("#download-detail").textContent = event.text || "model files and WebGPU runtime";
} else if (event.status === "initiate") {
$("#download-value").textContent = "cache check";
$("#download-detail").textContent = event.file;
}
}
function renderDirect(data) {
const output = $("#direct-output");
output.classList.remove("empty");
output.replaceChildren(...data.options.map((item) => {
const row = document.createElement("div");
row.className = "choice";
const label = document.createElement("span");
label.className = "choice-label";
const letter = document.createElement("b");
letter.textContent = item.label;
const description = document.createElement("small");
description.textContent = item.description;
label.append(letter, description);
const bar = document.createElement("span");
bar.className = "bar";
const fill = document.createElement("i");
fill.style.width = `${Math.max(1, item.probability * 100)}%`;
bar.append(fill);
const score = document.createElement("em");
score.textContent = item.probability.toFixed(3);
row.append(label, bar, score);
return row;
}));
$("#direct-total").textContent = seconds(data.totalMs);
$("#direct-input").textContent = `${data.inputTokens} tok`;
$("#direct-readouts").textContent = `${data.readouts} readout${data.readouts === 1 ? "" : "s"}`;
}
function resetResults() {
$("#direct-output").textContent = "running one forward pass…";
$("#direct-output").className = "output empty";
$("#generated-output").textContent = "waiting for direct readout…";
$("#generated-output").className = "output empty";
for (const id of ["#direct-total", "#direct-input", "#generation-ttft", "#generation-total", "#generation-input", "#generation-tokens"]) $(id).textContent = "—";
$("#direct-readouts").textContent = "—";
$("#ratio").textContent = "measuring…";
}
worker.addEventListener("message", ({ data }) => {
switch (data.type) {
case "progress":
renderProgress(data.event);
break;
case "loading":
setSupport(data.message);
break;
case "loaded":
$("#load-value").textContent = seconds(data.loadMs);
$("#download-meter").style.width = "100%";
if (!files.size) {
$("#download-value").textContent = "cached";
$("#download-detail").textContent = "no network transfer observed";
}
break;
case "ready":
ready = true;
$("#warmup-value").textContent = seconds(data.warmupMs);
setSupport(`Ready. ${data.modelName} is loaded locally on WebGPU.`, "ok");
loadButton.disabled = true;
modelSelect.disabled = true;
loadButton.innerHTML = '<span class="material-symbols-rounded" aria-hidden="true">check</span> model ready';
runButton.disabled = false;
break;
case "direct":
renderDirect(data);
$("#generated-output").textContent = "reading the decision…";
break;
case "generation-start":
$("#generated-output").textContent = "";
$("#generated-output").classList.remove("empty");
break;
case "generation-update":
$("#generated-output").textContent = data.text;
if (data.ttftMs != null) $("#generation-ttft").textContent = seconds(data.ttftMs);
$("#generation-tokens").textContent = `${data.tokens} tok`;
break;
case "complete": {
$("#generated-output").textContent = data.generatedText || "(no visible text generated)";
$("#generation-ttft").textContent = data.ttftMs == null ? "no token" : seconds(data.ttftMs);
$("#generation-total").textContent = seconds(data.generationMs);
$("#generation-input").textContent = `${data.inputTokens} tok`;
$("#generation-tokens").textContent = `${data.generatedTokens} tok`;
$("#ratio").textContent = `${(data.generationMs / data.directMs).toFixed(2)}× generation / direct`;
$("#run-note").textContent = `Measured sequentially in this tab. Direct: ${seconds(data.directMs)}. Generation: ${seconds(data.generationMs)}. Order is fixed and the model was warmed before both.`;
setSupport("Comparison complete. Edit the decision and run again whenever you like.", "ok");
runButton.disabled = false;
runButton.innerHTML = '<span class="material-symbols-rounded" aria-hidden="true">replay</span> run again';
break;
}
case "error":
setSupport(data.message, "error");
runButton.disabled = !ready;
loadButton.disabled = ready;
modelSelect.disabled = ready;
loadButton.innerHTML = ready
? '<span class="material-symbols-rounded" aria-hidden="true">check</span> model ready'
: '<span class="material-symbols-rounded" aria-hidden="true">refresh</span> retry model load';
break;
}
});
worker.addEventListener("error", (event) => {
setSupport(`Worker failed: ${event.message}`, "error");
loadButton.disabled = false;
});
function optionRows() {
return [...optionList.querySelectorAll(".option-row")];
}
function syncOptionControls() {
const rows = optionRows();
rows.forEach((row, index) => { row.querySelector("b").textContent = String.fromCharCode(65 + index); });
$("#option-count").textContent = `${rows.length} / ${MAX_OPTIONS}`;
removeOptionButton.disabled = rows.length <= MIN_OPTIONS;
addOptionButton.disabled = rows.length >= MAX_OPTIONS;
}
function appendOption(value = "") {
if (optionRows().length >= MAX_OPTIONS) return;
const row = document.createElement("label");
row.className = "option-row";
const label = document.createElement("b");
const input = document.createElement("input");
input.className = "option";
input.value = value;
input.placeholder = "Describe this option";
row.append(label, input);
optionList.append(row);
syncOptionControls();
return input;
}
function setOptions(values) {
while (optionRows().length > values.length) optionRows().at(-1).remove();
while (optionRows().length < values.length) appendOption();
optionRows().forEach((row, index) => { row.querySelector(".option").value = values[index]; });
syncOptionControls();
}
addOptionButton.addEventListener("click", () => appendOption()?.focus());
removeOptionButton.addEventListener("click", () => {
const rows = optionRows();
if (rows.length > MIN_OPTIONS) rows.at(-1).remove();
syncOptionControls();
});
syncOptionControls();
async function checkWebGPU() {
if (!navigator.gpu) {
setSupport("WebGPU is unavailable. Use a current WebGPU-capable browser over HTTPS or localhost.", "error");
loadButton.disabled = true;
return;
}
const adapter = await navigator.gpu.requestAdapter();
if (!adapter) {
setSupport("WebGPU exists, but no GPU adapter is available in this browser.", "error");
loadButton.disabled = true;
return;
}
setSupport("WebGPU is ready. The model does not download until you click load.", "ok");
}
loadButton.addEventListener("click", () => {
loadButton.disabled = true;
modelSelect.disabled = true;
loadButton.innerHTML = '<span class="material-symbols-rounded spin" aria-hidden="true">progress_activity</span> loading…';
worker.postMessage({
type: "load",
modelId: modelSelect.value,
useLocal:
["127.0.0.1", "localhost"].includes(location.hostname) &&
new URLSearchParams(location.search).has("local"),
});
});
document.querySelectorAll("[data-preset]").forEach((button) => {
button.addEventListener("click", () => {
const preset = presets[button.dataset.preset];
if (!preset) return;
$("#state").value = preset.state;
$("#question").value = preset.question;
setOptions(preset.options);
$("#state").focus();
});
});
modelSelect.addEventListener("change", renderSelectedModel);
modelSelect.value = "minicpm5-2b";
$("#device-note").textContent = isMobileDevice
? "Phone or small device detected · MiniCPM5 2B is selected. Switch to Qwen3 0.6B in the model box if loading is too heavy."
: "Desktop detected · MiniCPM5 2B selected by default.";
renderSelectedModel();
runButton.addEventListener("click", () => {
const state = $("#state").value.trim();
const question = $("#question").value.trim();
const options = [...document.querySelectorAll(".option")].map((input) => input.value.trim());
if (!state || !question || options.some((option) => !option)) {
setSupport("State, question and every option must be nonempty.", "error");
return;
}
resetResults();
runButton.disabled = true;
runButton.innerHTML = '<span class="material-symbols-rounded spin" aria-hidden="true">progress_activity</span> running…';
setSupport("Running direct readout, then autoregressive generation…");
worker.postMessage({ type: "compare", data: { state, question, options } });
});
checkWebGPU().catch((error) => setSupport(`WebGPU check failed: ${error.message}`, "error"));
|