Llama 2 is a collection of open-source large language models (LLMs) developed by Meta, available in various sizes for commercial and research use.
Llama 2 is a series of large language models released by Meta, building upon the original LLaMA formula with significant technical enhancements. These include a 40% larger pretraining corpus, doubled context length to 4k, and the adoption of grouped-query attention. The models are available in base and fine-tuned chat variants, with parameter sizes ranging from 7 billion to 70 billion. Meta has invested heavily in data quality, training techniques, capabilities evaluation, and safety training, aiming to provide a powerful alternative to closed-source LLMs. The development of Llama 2 involved extensive benchmarking and a two-stage Reinforcement Learning from Human Feedback (RLHF) process, utilizing rejection sampling and Proximal Policy Optimization (PPO). A key innovation is the use of two separate reward models, one optimized for helpfulness and another for safety, to address the safety-helpfulness tradeoff. The models are designed for commercial use, with a license that requires a form for access and has a restriction for products with over 700 million monthly active users.
Integrations: HuggingFace
Platforms: Web
View full Llama 2 profile on Tools-Radar | Browse Coding & Developer Tools tools | Alternatives to Llama 2
Tools-Radar is a free directory of 10,000+ AI tools — discover, compare, and choose the right AI software for your needs. Visit tools-radar.com