Even view it as a simple probability model, you don’t need a million Python repos. The massive amount of English text + 1,000 repos + your codebase is very powe

Even view it as a simple probability model, you don’t need a million Python repos. The massive amount of English text + 1,000 repos + your codebase is very powerful. You can see this because you can make up a language, give it some examples and it’s surprisingly good.