Create visialqa.html
Browse files- visialqa.html +168 -0
visialqa.html
ADDED
|
@@ -0,0 +1,168 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<!DOCTYPE html>
|
| 2 |
+
<html>
|
| 3 |
+
<head>
|
| 4 |
+
<meta charset="UTF-8"/>
|
| 5 |
+
<meta name="viewport" content="width=device-width, initial-scale=1.0"/>
|
| 6 |
+
<script src="https://cdn.tailwindcss.com"></script>
|
| 7 |
+
<!-- polyfill for firefox + import maps -->
|
| 8 |
+
<script src="https://unpkg.com/[email protected]/dist/es-module-shims.js"></script>
|
| 9 |
+
<script type="importmap">
|
| 10 |
+
{
|
| 11 |
+
"imports": {
|
| 12 |
+
"@huggingface/inference": "https://cdn.jsdelivr.net/npm/@huggingface/[email protected]/+esm"
|
| 13 |
+
}
|
| 14 |
+
}
|
| 15 |
+
</script>
|
| 16 |
+
</head>
|
| 17 |
+
<body>
|
| 18 |
+
<form class="w-[90%] mx-auto pt-8" onsubmit="launch(); return false;">
|
| 19 |
+
<h1 class="text-3xl font-bold">
|
| 20 |
+
<span
|
| 21 |
+
class="bg-clip-text text-transparent bg-gradient-to-r from-pink-500 to-violet-500"
|
| 22 |
+
>
|
| 23 |
+
Document & visual question answering demo with
|
| 24 |
+
<a href="https://github.com/huggingface/huggingface.js">
|
| 25 |
+
<kbd>@huggingface/inference</kbd>
|
| 26 |
+
</a>
|
| 27 |
+
</span>
|
| 28 |
+
</h1>
|
| 29 |
+
|
| 30 |
+
<p class="mt-8">
|
| 31 |
+
First, input your token if you have one! Otherwise, you may encounter
|
| 32 |
+
rate limiting. You can create a token for free at
|
| 33 |
+
<a
|
| 34 |
+
target="_blank"
|
| 35 |
+
href="https://huggingface.co/settings/tokens"
|
| 36 |
+
class="underline text-blue-500"
|
| 37 |
+
>hf.co/settings/tokens</a
|
| 38 |
+
>
|
| 39 |
+
</p>
|
| 40 |
+
|
| 41 |
+
<input
|
| 42 |
+
type="text"
|
| 43 |
+
id="token"
|
| 44 |
+
class="rounded border-2 border-blue-500 shadow-md px-3 py-2 w-96 mt-6"
|
| 45 |
+
placeholder="token (optional)"
|
| 46 |
+
/>
|
| 47 |
+
|
| 48 |
+
<p class="mt-8">
|
| 49 |
+
Pick the model type and the model you want to run. Check out models for
|
| 50 |
+
<a
|
| 51 |
+
href="https://huggingface.co/tasks/document-question-answering"
|
| 52 |
+
class="underline text-blue-500"
|
| 53 |
+
target="_blank"
|
| 54 |
+
>
|
| 55 |
+
document</a
|
| 56 |
+
> and
|
| 57 |
+
<a
|
| 58 |
+
href="https://huggingface.co/tasks/visual-question-answering"
|
| 59 |
+
class="underline text-blue-500"
|
| 60 |
+
target="_blank"
|
| 61 |
+
>image</a> question answering.
|
| 62 |
+
</p>
|
| 63 |
+
|
| 64 |
+
<div class="space-x-2 flex text-sm mt-8">
|
| 65 |
+
<label>
|
| 66 |
+
<input class="sr-only peer" name="type" type="radio" value="document" onclick="update_model(this.value)" checked />
|
| 67 |
+
<div class="px-3 py-3 rounded-lg shadow-md flex items-center justify-center text-slate-700 bg-gradient-to-r peer-checked:font-semibold peer-checked:from-pink-500 peer-checked:to-violet-500 peer-checked:text-white">
|
| 68 |
+
Document
|
| 69 |
+
</div>
|
| 70 |
+
</label>
|
| 71 |
+
<label>
|
| 72 |
+
<input class="sr-only peer" name="type" type="radio" value="image" onclick="update_model(this.value)" />
|
| 73 |
+
<div class="px-3 py-3 rounded-lg shadow-md flex items-center justify-center text-slate-700 bg-gradient-to-r peer-checked:font-semibold peer-checked:from-pink-500 peer-checked:to-violet-500 peer-checked:text-white">
|
| 74 |
+
Image
|
| 75 |
+
</div>
|
| 76 |
+
</label>
|
| 77 |
+
</div>
|
| 78 |
+
|
| 79 |
+
<input
|
| 80 |
+
id="model"
|
| 81 |
+
class="rounded border-2 border-blue-500 shadow-md px-3 py-2 w-96 mt-6"
|
| 82 |
+
value="impira/layoutlm-document-qa"
|
| 83 |
+
required
|
| 84 |
+
/>
|
| 85 |
+
|
| 86 |
+
<p class="mt-8">The input image</p>
|
| 87 |
+
|
| 88 |
+
<input type="file" required accept="image/*"
|
| 89 |
+
class="rounded border-blue-500 shadow-md px-3 py-2 w-96 mt-6 block"
|
| 90 |
+
rows="5"
|
| 91 |
+
id="image"
|
| 92 |
+
/>
|
| 93 |
+
|
| 94 |
+
<p class="mt-8">The question</p>
|
| 95 |
+
|
| 96 |
+
<input
|
| 97 |
+
type="text"
|
| 98 |
+
id="question"
|
| 99 |
+
class="rounded border-2 border-blue-500 shadow-md px-3 py-2 w-96 mt-6"
|
| 100 |
+
required
|
| 101 |
+
/>
|
| 102 |
+
|
| 103 |
+
<button
|
| 104 |
+
id="submit"
|
| 105 |
+
class="my-8 bg-green-500 rounded py-3 px-5 text-white shadow-md disabled:bg-slate-300"
|
| 106 |
+
>
|
| 107 |
+
Run
|
| 108 |
+
</button>
|
| 109 |
+
|
| 110 |
+
<p class="text-gray-400 text-sm">Output logs</p>
|
| 111 |
+
<div id="logs" class="bg-gray-100 rounded p-3 mb-8 text-sm">
|
| 112 |
+
Output will be here
|
| 113 |
+
</div>
|
| 114 |
+
|
| 115 |
+
<p>Check out the <a class="underline text-blue-500"
|
| 116 |
+
href="https://huggingface.co/spaces/huggingfacejs/doc-vis-qa/blob/main/index.html"
|
| 117 |
+
target="_blank">source code</a></p>
|
| 118 |
+
</form>
|
| 119 |
+
|
| 120 |
+
<script type="module">
|
| 121 |
+
import {HfInference} from "@huggingface/inference";
|
| 122 |
+
const default_models = {
|
| 123 |
+
"document": "impira/layoutlm-document-qa",
|
| 124 |
+
"image": "dandelin/vilt-b32-finetuned-vqa",
|
| 125 |
+
};
|
| 126 |
+
let running = false;
|
| 127 |
+
async function launch() {
|
| 128 |
+
if (running) {
|
| 129 |
+
return;
|
| 130 |
+
}
|
| 131 |
+
running = true;
|
| 132 |
+
try {
|
| 133 |
+
const hf = new HfInference(
|
| 134 |
+
document.getElementById("token").value.trim() || undefined
|
| 135 |
+
);
|
| 136 |
+
const model = document.getElementById("model").value.trim();
|
| 137 |
+
const model_type = document.querySelector("[name=type]:checked").value;
|
| 138 |
+
const image = document.getElementById("image").files[0];
|
| 139 |
+
const question = document.getElementById("question").value.trim();
|
| 140 |
+
document.getElementById("logs").textContent = "";
|
| 141 |
+
const method = model_type === "document" ? hf.documentQuestionAnswering : hf.visualQuestionAnswering;
|
| 142 |
+
const {answer, score} = await method({model, inputs: {
|
| 143 |
+
image, question
|
| 144 |
+
}});
|
| 145 |
+
document.getElementById("logs").textContent = answer + ": " + score;
|
| 146 |
+
} catch (err) {
|
| 147 |
+
alert("Error: " + err.message);
|
| 148 |
+
} finally {
|
| 149 |
+
running = false;
|
| 150 |
+
}
|
| 151 |
+
}
|
| 152 |
+
window.launch = launch;
|
| 153 |
+
window.update_model = (model_type) => {
|
| 154 |
+
const model_input = document.getElementById("model");
|
| 155 |
+
const cur_model = model_input.value.trim();
|
| 156 |
+
let new_model = "";
|
| 157 |
+
if (
|
| 158 |
+
model_type === "document" && cur_model === default_models["image"]
|
| 159 |
+
|| model_type === "image" && cur_model === default_models["document"]
|
| 160 |
+
|| cur_model === ""
|
| 161 |
+
) {
|
| 162 |
+
new_model = default_models[model_type];
|
| 163 |
+
}
|
| 164 |
+
model_input.value = new_model;
|
| 165 |
+
};
|
| 166 |
+
</script>
|
| 167 |
+
</body>
|
| 168 |
+
</html>
|