🚀 We're proud to unveil EXAONE 3.5 with three distinct models! Since the release of EXAONE 3.0, we've been actively listening to valuable feedback from LG Affiliates, companies, institutions, and academia. One consistent request was the need for models of varying sizes to accommodate different use cases - and we delivered. 💡 We've placed enterprise requirements at the forefront of our development, concentrating on fundamental and essential capabilities. We've prioritized achievable goals and ensured performance levels that genuinely enhance human productivity and operational efficiency in real industrial settings. 🎯 All models are developed through a responsible process, including AI ethical impact assessments, ensuring responsible innovation. This commitment to ethical AI development is at the core of our mission to create technology that benefits society while maintaining the highest standards of responsibility. 🤝 Looking to connect with partners who want to be part of our next success story. Reach out anytime!
📣We are open-sourcing #EXAONE 3.5, our #FrontierLevel models! We're so excited to see how you use them and welcome your valuable feedback! ☑️ An ultra-lightweight 2.4B model designed for on-device use. ☑️ A lightweight 7.8B model built for versatile applications. ☑️ A high-performance 32B model at the level of frontier models. All three models feature industry-leading #InstructionFollowing capabilities and #LongContext performance, outperforming global models of similar sizes. These models are ideal for applications aligned with recent research trends such as #RAG (Retrieval-Augmented Generation). ☑️ More Information on Our Blog : https://lnkd.in/gBBxEXBn ☑️ Model : https://lnkd.in/gsQzvj7B ☑️ Technical Report : https://lnkd.in/gckzTeHj #LG_AI_Research #LG_AI #AI