The similarities are way way too terrific to disregard. They probably experienced the model on the synthetic dataset produced by GPT-4o. Did Higher-Flyer misrepresent its utilization of GPUs to create DeepSeek seem to be extra economical than it really is? Was DeepSeek’s sudden public launch timed to push down Nvidia’s https://x.com/kidtsang/status/1884008035535782292