# GEMINI NANO EXTRACTED FROM CHROME CANARY VERSION 128.0.6557.0
If you have the particular version(>128), you can access the model here on windows:
C:\Users\USERNAME\AppData\Local\Google\Chrome SxS\User Data\OptGuideOnDeviceModel\[version]\weights.bin
You can signup for the built-in AI program here: https://developer.chrome.com/docs/ai/built-in
The model seems to be of the TFLite model format with the following architecture:
## Run the model using Mediapipe:
Instructions from here: https://github.com/google-ai-edge/mediapipe-samples/tree/main/examples/llm_inference/js
In a directory, make two files:
`index.html`:
```html
LLM Inference Web Demo
Input:
Result:
```
`index.js`(Replace line no. 23 to the path to the .bin file):
```js
// Copyright 2024 The MediaPipe Authors.
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
// http://www.apache.org/licenses/LICENSE-2.0
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
// ---------------------------------------------------------------------------------------- //
import {FilesetResolver, LlmInference} from 'https://cdn.jsdelivr.net/npm/@mediapipe/tasks-genai';
const input = document.getElementById('input');
const output = document.getElementById('output');
const submit = document.getElementById('submit');
const modelFileName = 'weights.bin'; /* PATH TO MODEL .bin */
/**
* Display newly generated partial results to the output text box.
*/
function displayPartialResults(partialResults, complete) {
output.textContent += partialResults;
if (complete) {
if (!output.textContent) {
output.textContent = 'Result is empty';
}
submit.disabled = false;
}
}
/**
* Main function to run LLM Inference.
*/
async function runDemo() {
const genaiFileset = await FilesetResolver.forGenAiTasks(
'https://cdn.jsdelivr.net/npm/@mediapipe/tasks-genai/wasm');
let llmInference;
submit.onclick = () => {
output.textContent = '';
submit.disabled = true;
llmInference.generateResponse(input.value, displayPartialResults);
};
submit.value = 'Loading the model...'
LlmInference
.createFromOptions(genaiFileset, {
baseOptions: {modelAssetPath: modelFileName},
// maxTokens: 512, // The maximum number of tokens (input tokens + output
// // tokens) the model handles.
// randomSeed: 1, // The random seed used during text generation.
// topK: 1, // The number of tokens the model considers at each step of
// // generation. Limits predictions to the top k most-probable
// // tokens. Setting randomSeed is required for this to make
// // effects.
// temperature:
// 1.0, // The amount of randomness introduced during generation.
// // Setting randomSeed is required for this to make effects.
})
.then(llm => {
llmInference = llm;
submit.disabled = false;
submit.value = 'Get Response'
})
.catch(() => {
alert('Failed to initialize the task.');
});
}
runDemo();
```
## Run using:
`python3 -m http.server 8000` in the same directory (or python -m SimpleHTTPServer 8000 for older python versions).
Finally,
`Open localhost:8000 in Chrome`
## License
I dont own shit. Just extracted the model weights stored as a .bin file. https://policies.google.com/terms/generative-ai/use-policy