Insert fed joke here... yeah, my understanding is that the resources needed to train models (both the data and hardware) are too massive for individuals to hope to undertake on their own. I kind of assumed that China putting out open source weights in past (since the big "deepseek moment a year ago...