I'm excited for open source code generation models. So I won't have to pay Github every month. And if this is a bigger dataset and permissively licensed, this means there will be no chance that it will generate copyrighted code.