| Gemini Offers Huge Context Window |
| Written by Kay Ewbank | |||
| Monday, 08 July 2024 | |||
|
Google has announced a range of improvements to Gemini, along with the release of Gemma 2. The first enhancement is access to a 2 million context window for Gemini 1.5 Pro, backed up with context caching to reduce overheads. Support for code execution capabilities in the Gemini API were also announced, along with the addition of Gemma 2 in Google AI Studio. Google has used the name Gemini both for its conversational chatbot previously known as Bard, and its multimodal large language model (LLM) developed by Google DeepMind. Gemini Pro refers to the LLM.
The longer context windows were first revealed at Google I/O, where Google announced the longest ever context window of 2 million tokens in Gemini 1.5 Pro behind a waitlist. That 2 million token context window on Gemini 1.5 Pro has now been opened for all developers. Tokens are the smallest building blocks, like part of a word, image or video that a model can process at once, and longer context windows are important because they help AI models recall information during a session. However, as the context window grows, so does the potential for input costhence the addition of support for context caching in the Gemini API for both Gemini 1.5 Pro and 1.5 Flash. Context caching reduces costs for tasks that use the same tokens across multiple prompts. In a typical AI workflow, you might pass the same input tokens mutliple times to a model. Context caching means you can pass some content to the model once, cache the input tokens, and then refer to the cached tokens for subsequent requests. A potentially more interesting addition for developers is the new support for code execution. This can be used to generate and run Python code and learn iteratively from the results until it gets to a desired final output. The execution sandbox is not connected to the internet, comes standard with a few numerical libraries, and developers are simply billed based on the output tokens from the model. The feature has been added to improve the handling of problems that require math or data reasoning analysis. Google has also announced that Gemini 1.5 Flash is now in production. Gemini 1.5 Flash aims to offer speed and affordability. The final part of the announcement was the addition of the open Gemma 2 model for experimentation in Google AI Studio. Gemma is Google's family of lightweight open models that have been built from the same research and technology used to create Google's Gemini models. The models in Gemma are text-to-text, decoder-only large language models, available in English, with open weights, pre-trained variants, and instruction-tuned variants. Gemini 1.5 Pro and Gemma 2 are available now in Google AI Studio and Gemini Advanced which you can access with a Google One AI Premium plan that comes free for 2 months and costs $19.99 after that.
More InformationRelated ArticlesGoogle Rebrands Bard As Gemini With Subscription Google Releases Gemma Open Models Google Adds Code Generation To Bard To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin. 
Comments
or email your comment to: comments@i-programmer.info |
