Now Reading
Falcon 2 by TII is better than Google and Meta?

Falcon 2 by TII is better than Google and Meta?

Falcon take off to overtake Google and Meta

The Technology Innovation Institute (TII), supported by the Abu Dhabi government, has introduced Falcon 2, its latest large language model, to rival counterparts from Meta, Google, and OpenAI. These advanced AI models emulate human intelligence, performing tasks like language translation and visual interpretation using extensive datasets.

As the applied research arm of Abu Dhabi’s Advanced Technology Research Council (ATRC), TII announced Falcon 2 in two versions: Falcon 2 11B and Falcon 2 11B VLM. These models are open-source, enabling developers worldwide to access and utilize them freely across various languages, including English, French, Spanish, German, and Portuguese.

Faisal Al Bannai, ATRC’s secretary-general, reiterated their commitment to open-source principles, aiming to empower developers and entities globally with access to cutting-edge AI technology. Falcon 2 11B demonstrates superior performance compared to Meta’s Llama 3 and matches Google’s Gemma 7B, according to assessments by Hugging Face, a platform for the machine learning community.

Additionally, Falcon 2 11B VLM boasts image interpretation capabilities, offering extensive applications across sectors like healthcare, finance, e-commerce, education, and law. These applications range from document management and digital archiving to assisting visually impaired individuals, enhancing accessibility and inclusivity.

Moreover, TII emphasizes the efficiency and scalability of Falcon 2 models, which can run seamlessly on a single graphics processing unit (GPU). This feature ensures easy deployment and integration into various infrastructures, including lightweight systems like laptops.

See Also
M42, UAE Defence Partner to Run Hospital

Looking ahead, TII plans to expand the Falcon 2 lineup by introducing a range of new sizes. These enhanced models will incorporate advanced machine learning capabilities such as a “mixture of experts.” This innovative approach involves merging smaller networks with distinct specializations to deliver highly complex and tailored responses, further enhancing the models’ performance and versatility.

About Author

© 2021 The Technology Express. All Rights Reserved.

Scroll To Top