Update README.md

README.md

@@ -12,4 +12,89 @@ license: apache-2.0
 
 https://huggingface.co/qnguyen3/nanoLLaVA with ONNX weights to be compatible with Transformers.js.
 
+## Usage (Transformers.js)
+
+> [!IMPORTANT]
+> NOTE: nanoLLaVA support is experimental and requires you to install Transformers.js [v3](https://github.com/xenova/transformers.js/tree/v3) from source.
+
+If you haven't already, you can install the [Transformers.js](https://huggingface.co/docs/transformers.js) JavaScript library from [GitHub](https://github.com/xenova/transformers.js/tree/v3) using:
+```bash
+npm install xenova/transformers.js#v3
+```
+
+**Example:**
+```js
+import { AutoProcessor, AutoTokenizer, LlavaForConditionalGeneration, RawImage } from '@xenova/transformers';
+
+// Load tokenizer, processor and model
+const model_id = 'Xenova/nanoLLaVA';
+const tokenizer = await AutoTokenizer.from_pretrained(model_id);
+const processor = await AutoProcessor.from_pretrained(model_id);
+const model = await LlavaForConditionalGeneration.from_pretrained(model_id, {
+    dtype: {
+        embed_tokens: 'fp16', // or 'fp32' or 'q8'
+        vision_encoder: 'fp16', // or 'fp32' or 'q8'
+        decoder_model_merged: 'q4', // or 'q8'
+    },
+    // device: 'webgpu',
+});
+
+// Prepare text inputs
+const prompt = 'What does the text say?';
+const messages = [
+    { role: 'system', content: 'Answer the question.' },
+    { role: 'user', content: `<image>\n${prompt}` }
+]
+const text = tokenizer.apply_chat_template(messages, { tokenize: false, add_generation_prompt: true });
+const text_inputs = tokenizer(text);
+
+// Prepare vision inputs
+const url = 'https://huggingface.co/qnguyen3/nanoLLaVA/resolve/main/example_1.png';
+const image = await RawImage.fromURL(url);
+const vision_inputs = await processor(image);
+
+// Generate response
+const { past_key_values, sequences } = await model.generate({
+    ...text_inputs,
+    ...vision_inputs,
+    do_sample: false,
+    max_new_tokens: 64,
+    return_dict_in_generate: true,
+});
+
+// Decode output
+const answer = tokenizer.decode(
+    sequences.slice(0, [text_inputs.input_ids.dims[1], null]),
+    { skip_special_tokens: true },
+);
+console.log(answer);
+// The text reads "Small but mighty".
+
+const new_messages = [
+    ...messages,
+    { role: 'assistant', content: answer },
+    { role: 'user', content: 'How does the text correlate to the context of the image?' }
+]
+const new_text = tokenizer.apply_chat_template(new_messages, { tokenize: false, add_generation_prompt: true });
+const new_text_inputs = tokenizer(new_text);
+
+// Generate another response
+const output = await model.generate({
+    ...new_text_inputs,
+    past_key_values,
+    do_sample: false,
+    max_new_tokens: 256,
+});
+const new_answer = tokenizer.decode(
+    output.slice(0, [new_text_inputs.input_ids.dims[1], null]),
+    { skip_special_tokens: true },
+);
+console.log(new_answer);
+// The context of the image is that of a playful and humorous illustration of a mouse holding a weightlifting bar. The text "Small but mighty" is a playful reference to the mouse's size and strength.
+```
+
+We also released an online demo, which you can try yourself: https://huggingface.co/spaces/Xenova/experimental-nanollava-webgpu
+
+<video controls autoplay src="https://cdn-uploads.huggingface.co/production/uploads/61b253b7ac5ecaae3d1efe0c/0T-aNjgXt6PGL3qIl8wBc.mp4"></video>
+
 Note: Having a separate repo for ONNX weights is intended to be a temporary solution until WebML gains more traction. If you would like to make your models web-ready, we recommend converting to ONNX using [🤗 Optimum](https://huggingface.co/docs/optimum/index) and structuring your repo like this one (with ONNX weights located in a subfolder named `onnx`).