The research that went into the flagship of the company is open-sourced and can be used more freely by developers. The lightweight Gemma is likely to be suitable for smaller tasks, as it is nearly as powerful as the big closed model, Gemini. The models don't have much in the way of complexity, but they may make up for it in speed and cost. Despite their smaller size, the models are able to run directly on a developer laptop or desktop computer, despite their larger size. The way that Gemini was released is different to the way that Gemma is released. More people will experiment with the benefits of the open source model instead of using competitors that offer better access. Many companies prohibit their models from being used for specific tasks, such as weapons development programs. The company undertook more extensive red-teaming to Gemma because of the inherent risks involved with open models, according to Tris Warkentin.
The tool will allow developers to create guidelines or banned word list when using the tool. Warkentin said that the models work best for English-language tasks. The company said researchers can apply for up to $500,000 in cloud credits. It is not clear how much of a demand there is for smaller models, but other companies have released lighter-weight versions of their flagship models. Business users and developers will be happy to know that Google recently announced a faster Gemini 1.5, for business users and developers.