Comment by hongbo_zhang

This is the benchmark between the latest models on a new programming language to avoid overfitting. Latest models are quite good over generalization to new languages, they can write tens of thousands of lines of code in one prompt that just works.