The problem _there_ is that there's very little diversity in the training data - it's all NVidia designs which are probably from the same phylogenetic tree. It'

The problem _there_ is that there's very little diversity in the training data - it's all NVidia designs which are probably from the same phylogenetic tree. It'll probably end up regurgitating existing NV designs...