According to PANews, Ahmad Al-Dahle, the leader of generative AI at Meta Platforms, addressed concerns regarding the recently open-sourced Llama 4 model in a detailed post on social media at 1:49 AM UTC+8. Ahmad explained that the model was released immediately after development, which may have led to variations in model quality across different services. Meta is working swiftly to address these issues and enhance performance. Additionally, Ahmad denied any pre-training on the test set
source: https://www.binance.com/en/square/post/22621702950169?utm_source=BinanceNewsRSS