Google’s functiongemma for quick undErstanding

Google just released a new model aimed at function calling, read their article below:

https://blog.google/technology/developers/functiongemma/

You may be looking at all this, seeing reposts from developers and huggingfaces and cant make heads or tails of what it actually is. Well fear not I will give you the VERY brief explanation.

FunctionGemma is a small model that can fit on something as small as your phone. Its specialty is in its name, it makes function calls or api calls and can handle a little bit of multi step calls.

When you see them have the games in the article they are tell FunctionGemma through voice in natural human language to perform actions. FunctionGemma is then taking that language translating the intent into functions using the apis it has access to. Then executing those functions it created to achieve what was asked of it in natural language.

If that was still too much for you here is the simplest way I can make it. It’s the Siri or Google assistant feature you always dreamed of. Currently they have set things they will do for you. But, this opens the pathway so that anything you could do on your phone, potentially it can do for you while you just tell it with your own words.

It is especially cool because it does not need to make any of these calls off device. It is a private and secure model for your own personal use.

There are many more cool applications that can be done to, I currently picture having a raspberry pi or something attached to a camera running FunctionGemma you can then have it do all sorts of cool custom functions you may not have been able to write yourself. Granted this example would also need a vision model running but I just came up with this off the cuff so give me some slack, “track the number of times and timestamp my dog goes outside”. Could save an audio clip on raspberrypi and “if an animal or person passes into view play sound clip” or if the device is internet enabled could be for security “open camera and video record once recording ends send video to my email”.

Those quick ones are just what I came up with. I would love to hear what you all can come up with!

Next
Next

Creating Auto DM (Anansi)