@@WL-mi9pl Politician is Politician, bussinessman is bussinessman, regardless of education backgrouds. It is true that in treal world that 文科生 is disadvantaged "over" 理工背景.
Congrats to you on your quick learning capability on those topics. I have a fews points: 1. Deepseek never revealed the actually costs. 5.5 miliiion is an estimattion by CNBC. It is hard to estimate the cost as I believe that they use both NV GPU and Huaei Ascend GPU. 2. The significant of DS is that it shows again that algorithm (not software) is much more important than hardware. A previous example is back-propagation algoriothm, which enabled the application of learning based on NN. 3. The importance of CUDA is over-exaggerated as it is easily bypassed by DS engineers. I am able to code in CUDA after half day training. I am an enbgineer, not a computer graduate. People wishfully exaggerate CUDA in the same way TW people do to TSMC.