We all know how it has been a few months since Google unveiled Gemini, and the company even recently launched a new update. Now, the search engine giant has taken the responsibility of launching Gemma; this language model is aimed at developers and is going to help them build AI responsibly. It is now available all over the world in 2B and 7B. Both models are pre-trained and instruction-tuned variants. Google claims that it performs better in MMLU benchmarks than Mistral 7B and Llama 13B.
Google's Gemma is a way forward for developers who are looking to build AI-based applications responsibly. It's free to use and can run on the majority of machines without specialized hardware
One of the best things about Gemma compared to some of the other models is that it can run on computers and laptops without any issues. Google even talks about how it excels at some key benchmarks where it can deliver better performance than the alternative models. As expected, this model is built on Google's AI principles, and the company made sure to test it a lot to make sure that it does not give responses that are not in line with the principles Google has put in place.
In addition to releasing Gemma, Google also released a Responsible Generative AI Toolkit that will help developers and researchers build safe AI applications. Here is a rundown of the new toolkit.
- Safety classification: A novel methodology for building robust safety classifiers with minimal examples.
- Debugging: A model debugging tool helps you investigate Gemma's behavior and address potential issues.
- Guidance: You can access best practices for model builders based on Google’s experience in developing and deploying large language models.
Moving forward, Gemma also supports the following tools and systems so you can have an easier time understanding what you are getting into.
- Multi-framework tools: There are reference implementations for inference and fine-tuning across multi-framework Keras 3.0, native PyTorch, JAX, and Hugging Face Transformers.
- Cross-device compatibility: Gemma models run across many different device types, including laptop, desktop, IoT, mobile, and cloud.
- Nvidia partnership: Gemma is optimized for Nvidia GPUs, thanks to a partnership between Google and Nvidia.
- Optimized for Google Cloud: Vertex AI provides a broad MLOps toolset with a range of tuning options and one-click deployment using built-in inference optimizations. Advanced customization is available with fully-managed Vertex AI tools or with self-managed GKE, including deployment to cost-efficient infrastructure across GPU, TPU, and CPU from either platform.
The good news is that Gemma is free to use on Kaggle, a free tier for Colab notebooks, and you also get $300 in credits as a first-time Google Cloud user. The company is even helping researchers by letting them apply for up to $500,000 in free Google Cloud credits.
The best thing is that Gemma is free to try, as you will be able to run it on most modern machines with enough vRAm and an Nvidia GPU. You can give it a try here if you think the tool is something that you need.
We have criticized Google time and again for not taking its products and services seriously. But with the recent update to Gemini and the release of Gemma, it has become evident that the company wants to do right by everyone who is looking to excel in the field of artificial intelligence. I honestly cannot wait to see what the future holds for all things AI.
News Source: Google Blog
About the author: I have been tinkering with Android devices ever since the early days of the HTC Desire. Over time, I have grown a fondness for the ecosystem and now I cannot live without it. Although some might believe that I have sold my soul to Android, but I believe it is not the case. You can find me writing tutorials and posting guides on a number of different smartphones. When I am not writing here, I am wasting myself away in books, journals, or on Steam.
Follow Wccftech on Google to get more of our news coverage in your feeds.
