The open source brigade of artificial intelligence’s large language models (LLM) is growing, with over 65.7 per cent of all new models being released for free access in 2023 following the open source route. However, despite strength in numbers, they are not necessarily performing the best. Closed-source models outperformed their open-sourced counterparts on 10 select benchmarks, achieving a median performance advantage of 24.2 per cent, a new study by Stanford Institute said. ET collates the key... Read this story
