Meta and Booz Allen Hamilton have initiated the deployment of Meta's open-source AI model, Llama, in space. Astronauts on the International Space Station (ISS) will use a fine-tuned version, Llama 3.2, for scientific research. This technology includes Meta's Llama 3.2, supported by Hewlett Packard Enterprise's Spaceborne Computer-2 and Nvidia's GPU.
The Space Llama project aims to enhance research capabilities on the ISS. According to Meta's Vice President Ahmad Al-Dahle, open-source AI models like Llama are expected to play a crucial role in advancing space exploration and research, enabling astronauts to conduct complex experiments and drive innovation.
Booz Allen Hamilton's announcement builds on their previous ISS initiatives, including the deployment of the first large language model in space last August. The new technology combination, Space Llama, demonstrates how large language models can function in challenging environments with limited bandwidth and intermittent connectivity.
The project is designed to reduce costs, minimize computing power needs, and facilitate faster issue resolution without relying on Earth-based internet. Space Llama will help astronauts replace paper documents and lessen dependence on ground instructions, streamlining operations in space.
