Meta partners with Microsoft, unveils open-source AI model 'Llama 2'

Meta and Microsoft released Llama 2, an open-source large language model available for both research and commercial use, marking a significant shift from the previous non-commercial LLaMA release. By making these powerful tools accessible across major cloud platforms and locally on Windows, Meta aims to accelerate innovation and enable developers and businesses to build their own generative AI applications without restrictive licensing barriers. While Llama 2 demonstrates competitive capabilities against other open-source models, it currently trails behind top proprietary systems like GPT-4, particularly in coding proficiency. Despite this performance gap, the model represents a crucial step toward democratizing advanced AI technology, allowing a broader ecosystem of creators to experiment with and refine large language models without needing extensive computational resources or exclusive partnerships. This release is highly relevant to open_data because it highlights ongoing tensions between open development and data transparency. Although Meta provided detailed methodology for training and safety evaluations, the lack of clear definition regarding training data sources remains a critical concern for the community. This underscores the importance of complete data provenance in open-source AI, ensuring that ethical standards and reproducibility can be rigorously maintained as these technologies become more widespread.

Source: medianama.com
Published on 2023-07-21

Las ventas de las empresas caen un 2,5% en mayo y acumulan dos meses de bajadas »

Open Data

Related news