Does it have performance on the code evaluation dataset

[Guncuke]

Great job! But I noticed that many examples were demonstrated by whether the code could compile successfully or if there were incomplete snippets. However, the experiments seem to be based on math evaluation sets. Any results on code evaluation sets like LiveCodeBench?

[ltzheng]

Thanks for your interest! The code reward in this repo is experimental. We have an ongoing project on code tasks and will release it soon.