In this guide, we will show you how to easily install the new Google Gemma 3N on Android Mobile. The GMMA 3N can be turned on a 2 GB RAM. Participates in the same architecture as Gemini Nano and is designed for incredible performance. Google added an audio understanding, so it is now multimedia, fast and effective to resources, and operating it on the device – it is not necessary to have a cloud connection!
GEMMA 3N It is a new model of prosthesis from Google that provides strong and effective features directly on devices such as mobile phones, tablets and laptops. One can also test GEMMA 3N in Google to study.
Size variablesAvailable in four different sizes (1B, 4B, 12B and 27B) for optimal flexibility.
To implement the installation and you will need two files: APK APK and GEMMA-3N-E2B-ETERT-PREVIEW This is Edge.
1- Download Edge Gallery APK:
Download the file from GitHub here AI-Edge-Gallery.APK. Then click Install its installation.
2- Download one of the .task files from the embrace of the embrace:
download Gemma-3N-E2B or E4B version. You will need an embraced account and need to agree to the conditions and conditions.
4- Start the Edge Gallery app and add the GMMA 3N file:
Open the Edge Gallery app and click In addition to the symbol at the bottom right Add the Gemma 3N file to our case is E2B. Choose whether GEMMA should run on GPU or the CPU.
4- You can make a chat with GEMMA 3N. Amnesty International Chat
Click on AI chat and then choose at the bottom of GEMMA..T4
Tip: Change the tone on LLM chat.
To change the chat response format, go to the claim laboratory and adjust the Gemma conversation text (for example, from the formal style to the informal style).
More information about GEMMA 3 and features:
- The model is impressive for its size and can perform OCR and interpret the image locally, but it has limited knowledge and not perfect for complex tasks.
- Language SupportSupports more than 35 languages directly and can handle more than 140 languages in total.
- Multimedia support: Pictures, texts and short videos can be analyzed.
- Window on expanded contextIt provides a 128K symbolic context window to process large amounts of information.
- Functional: Supports job calls and organized output of data automation.
- ImprovedIt includes the quantitative measurement to reduce the size of the model without losing accuracy.
