AI2's Release of OLMo 2 Language Models: A Fully Open Competitor To Llama

The Allen Institute for AI (AI2) has recently launched OLMo 2, a new family of fully open-source language models. This release is significant in the field of artificial intelligence (AI) as it represents a major advancement in open-source AI technology, challenging proprietary models like Meta's Llama series.
Development and Technical Specifications
OLMo 2 consists of two primary models: OLMo 7B and OLMo 13B. These models are named after their parameter sizes, with 7 billion and 13 billion parameters, respectively. The development of OLMo 2 was guided by AI2's commitment to transparency and reproducibility, aligning with the Open Source Initiative's definition of open-source AI.