Update index.js

index.js

@@ -1,4 +1,6 @@
-import { pipeline, env, AutoTokenizer, RawImage, Tensor, getModelJSON  } from 'https://cdn.jsdelivr.net/npm/@huggingface/transformers';
+import { env, AutoTokenizer, RawImage } from 'https://cdn.jsdelivr.net/npm/@huggingface/transformers';
+import { getModelJSON } from "https://cdn.jsdelivr.net/npm/@huggingface/transformers@3.0.2/src/utils/hub.js";
+import { Tensor } from "https://cdn.jsdelivr.net/npm/@huggingface/transformer/utils/tensor.js";
 import * as ort from "https://cdn.jsdelivr.net/npm/onnxruntime-web@1.20.0/dist/ort.webgpu.mjs";
 
 // Since we will download the model from the Hugging Face Hub, we can skip the local model check
@@ -10,8 +12,8 @@ const fileUpload = document.getElementById('upload');
 const imageContainer = document.getElementById('container');
 const example = document.getElementById('example');
 
-const EXAMPLE_URL = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/car.jpg";
-const INPUT_IMAGE_SIZE = [960, 960];
+const EXAMPLE_URL = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/car.jpg",
+const INPUT_IMAGE_SIZE = [960, 960] as const;
 const HEIGHT_FACTOR = 10;
 const WIDTH_FACTOR = 10;
 const IMAGE_EMBED_SIZE = WIDTH_FACTOR * HEIGHT_FACTOR;
@@ -59,14 +61,14 @@ async function detect(img) {
 }
 
 
-export async function imageTextToText(
-  imagePath,
-  query,
+export async function simplifiedLLMVision(
+  imagePath: string,
+  query: string,
   vision = true
 ) {
   const suffix = QUANTIZATION ? `_${QUANTIZATION}` : "";
 
-  const config = (await getModelJSON(BASE_MODEL, "config.json"));
+  const config = (await getModelJSON(BASE_MODEL, "config.json")) as any;
 
   const prompt_head_len = new Tensor("int64", new BigInt64Array([5n]), [1]);