A Target recently announced the release of the model intelligence artificial (AI) open source is called image anchor, the big advantage of which is the management of various data streams. At the moment, it is still under study and has no practical application. instantbut it could be a big step forward for the technology — and for the company itself, given the difficulties it had with LLaMA.

The goal of this AI is to be multi-sensor and serve up six “sectors”: text, audio, visual, depth, temperature and motion readings in a single index that provides a cross-reference of data for machine learning. According to Meta, ImageBind is the first model to combine this many types.

The company believes that in the future, artificial intelligence systems will be able to create more and more immersive experiences and integrate multiple areas at the same time, simulating real situations in various formats.

Currently, it is already possible to visually place the user in different “worlds”, for example, on a roller coaster or in another room. The idea is to improve other sensors and include environmental sensations and even more movements.

Open source to “protect” yourself from leaks

Another interesting aspect of the launch is that Meta exposes model code, which is an increasingly common practice in the AI ​​world – and perhaps a way for Meta to “protect” itself from leaks since LLaMA, which was its version ChatGPTa leak on the developer forums.

While it has yet to achieve commercialization in AI and has seen competitors such as Microsoft And bing, successful, your strategy is a way to keep you relevant in the market. It is not yet known what the practical applications of ImageBind will be, but it should be well “tested” by developers in this area and therefore improved.

Mundo Conectado Deal Center: Selection of Discounts and Lowest Prices
Best deals on electronics, cell phones, TVs, soundbars, drones and more

…..