Oregon should use Yann LeCun's JEPA architecture for football tape analysis. A world model predicting plays over pixels gives the Ducks an edge! @oregonfootball@CoachDanLanning@ylecun
@oregonfootball@CoachDanLanning@ylecun JEPA is not a plug-and-play football AI yet, but its architecture is fundamentally suitable for learning strategic defensive insights from raw game video. With proper training and interface design, it could become a transformative tool for sports analytics.
🧵 College football isn't won just in the weight room anymore they’re won in the film room. It is time for @TexasTechFB and @DataScienceTTU to invest capital into cutting-edge AI architecture. Traditional analytics are hitting a ceiling. The sport demands world models. @ylecun
@TexasTechFB@DataScienceTTU@ylecun The ultimate goal in Lubbock is not just to compete—it is to hang banners. Adopting advanced machine learning intelligence like JEPA provides the exact structural advantage needed to dominate the conference and bring a National Championship home to Texas Tech.
@JPWeiland What if the R0 was 0.5 instead of 1.5, but still a GI of 20 days, what would the infected population be in the USA after 1 year? And what if it started the year (day 1) with 3 people (Georgia, Arizona, California)?
Remember RedNote, the TikTok alternative that had a surge of popularity in Jan?
Even they have an AI lab training open-source LLMs !!
Their 142B-A14B MoE model achieves performance comparable to Qwen2.5-72B