Update index.js
Browse files
index.js
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
import { env, AutoTokenizer, RawImage, Tensor, getSession } from 'https://cdn.jsdelivr.net/npm/@huggingface/transformers';
|
2 |
-
import { getModelJSON } from "https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.0.2/src/utils/hub.js";
|
3 |
import * as ort from "https://cdn.jsdelivr.net/npm/onnxruntime-web@1.20.0/dist/ort.webgpu.mjs";
|
4 |
|
5 |
// Since we will download the model from the Hugging Face Hub, we can skip the local model check
|
@@ -19,12 +19,10 @@ const IMAGE_EMBED_SIZE = WIDTH_FACTOR * HEIGHT_FACTOR;
|
|
19 |
const MAX_SEQ_LENGTH = 1024;
|
20 |
const ONNX_URL = "http://localhost:3004/onnx";
|
21 |
const BASE_MODEL = "Qwen/Qwen2-VL-2B-Instruct";
|
22 |
-
const
|
|
|
23 |
const MAX_SINGLE_CHAT_LENGTH = 10;
|
24 |
|
25 |
-
// const ONNX_MODEL_BASE_URL=
|
26 |
-
console.log(getSession);
|
27 |
-
|
28 |
status.textContent = 'Loading model...';
|
29 |
status.textContent = 'Ready';
|
30 |
|
@@ -47,7 +45,6 @@ fileUpload.addEventListener('change', function (e) {
|
|
47 |
reader.readAsDataURL(file);
|
48 |
});
|
49 |
|
50 |
-
|
51 |
async function parse(img, txt) {
|
52 |
imageContainer.innerHTML = '';
|
53 |
imageContainer.style.backgroundImage = `url(${img})`;
|
@@ -127,7 +124,7 @@ export async function imageTextToText(
|
|
127 |
const pixel_values = image.unsqueeze(0);
|
128 |
|
129 |
const ortSessionA = await ort.InferenceSession.create(
|
130 |
-
|
131 |
{ executionProviders: ["webgpu"] }
|
132 |
);
|
133 |
|
|
|
1 |
import { env, AutoTokenizer, RawImage, Tensor, getSession } from 'https://cdn.jsdelivr.net/npm/@huggingface/transformers';
|
2 |
+
import { getModelJSON, getModelFile } from "https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.0.2/src/utils/hub.js";
|
3 |
import * as ort from "https://cdn.jsdelivr.net/npm/onnxruntime-web@1.20.0/dist/ort.webgpu.mjs";
|
4 |
|
5 |
// Since we will download the model from the Hugging Face Hub, we can skip the local model check
|
|
|
19 |
const MAX_SEQ_LENGTH = 1024;
|
20 |
const ONNX_URL = "http://localhost:3004/onnx";
|
21 |
const BASE_MODEL = "Qwen/Qwen2-VL-2B-Instruct";
|
22 |
+
const ONNX_MODEL = "pdufour/Qwen2-VL-2B-Instruct-ONNX-Q4-F16";
|
23 |
+
const QUANT = "q4f16";
|
24 |
const MAX_SINGLE_CHAT_LENGTH = 10;
|
25 |
|
|
|
|
|
|
|
26 |
status.textContent = 'Loading model...';
|
27 |
status.textContent = 'Ready';
|
28 |
|
|
|
45 |
reader.readAsDataURL(file);
|
46 |
});
|
47 |
|
|
|
48 |
async function parse(img, txt) {
|
49 |
imageContainer.innerHTML = '';
|
50 |
imageContainer.style.backgroundImage = `url(${img})`;
|
|
|
124 |
const pixel_values = image.unsqueeze(0);
|
125 |
|
126 |
const ortSessionA = await ort.InferenceSession.create(
|
127 |
+
await getModelFile(ONNX_MODEL, "onnx/QwenVL_A_${QUANT}.onnx"),
|
128 |
{ executionProviders: ["webgpu"] }
|
129 |
);
|
130 |
|