Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,39 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# 🤖 Gemma TFLite Models for Android Demos
|
| 6 |
+
|
| 7 |
+
This repository hosts quantized TensorFlow Lite (`.tflite`) versions of Google's Gemma models, specifically for use in Android sample applications, hackathons, and educational projects.
|
| 8 |
+
|
| 9 |
+
The goal is to provide a frictionless, anonymous download path for on-device integration without requiring app users to log in or developers to embed secret API keys.
|
| 10 |
+
|
| 11 |
+
## 🤔 Why This Repository Exists
|
| 12 |
+
|
| 13 |
+
The original Gemma models are created by Google and are officially hosted on platforms like [Kaggle](https://www.kaggle.com/models/google/gemma-3n) and the [official Google Hugging Face account](https://huggingface.co/google).
|
| 14 |
+
|
| 15 |
+
These official repositories are often **"gated,"** meaning they require a user to be logged in and to have explicitly accepted the license terms before a download can start. This is a necessary compliance step for Google to ensure users agree to their Responsible AI terms.
|
| 16 |
+
|
| 17 |
+
However, this authentication requirement prevents direct, anonymous downloads from within a mobile application and would create a poor user experience (e.g., asking an app user to log in to a developer service).
|
| 18 |
+
|
| 19 |
+
This repository acts as a public, non-gated mirror. The models are redistributed here under the permissions granted by their **Apache 2.0 License**. This allows mobile demo apps to fetch the model file via a direct URL without any user-side authentication, while still respecting the original license terms.
|
| 20 |
+
|
| 21 |
+
## 📱 Intended Use
|
| 22 |
+
|
| 23 |
+
The models here are intended for developers who need to quickly integrate a Gemma model into an Android application for demonstration or prototyping purposes.
|
| 24 |
+
|
| 25 |
+
They can be downloaded directly within an app using Android's `DownloadManager` or a networking library like OkHttp, providing a seamless "first-run" experience for the end-user.
|
| 26 |
+
|
| 27 |
+
## 📦 Models Included
|
| 28 |
+
|
| 29 |
+
* `gemma-3n-e2b-it-int4.task`
|
| 30 |
+
* `gemma-3n-e4b-it-int4.task`
|
| 31 |
+
* *(More models may be added)*
|
| 32 |
+
|
| 33 |
+
## 📜 License and Attribution
|
| 34 |
+
|
| 35 |
+
All models in this repository are the property of Google and are governed by the **Apache 2.0 License** and the **Gemma Terms of Use**.
|
| 36 |
+
|
| 37 |
+
**Official Gemma Terms of Use:** [https://ai.google.dev/gemma/terms](https://ai.google.dev/gemma/terms)
|
| 38 |
+
|
| 39 |
+
**⚠️ IMPORTANT:** If you use these models in your application, you **must include proper attribution to Google** and provide a copy of or link to the license in your app, as required by the license terms.
|