PRM Training Data

[SetonLiang]

Hi, I noticed that the provided BIRD train dataset contains 18,015 samples, which seem to be from the "After Syntax Tree Deduplication Stage". I would like to ask: is this the dataset used for training?

Or should we use the final "Filtered PRM Training Data" instead? If so, could you please provide that dataset and its format? Thanks a lot!

[facdbe-kt]

Thank you for your question about the dataset!

Currently, we haven't provided the "Filtered PRM Training Data" yet. However, we plan to include this data in our next release, which will be available soon in both zip format and on Hugging Face.

We appreciate your interest in our work and thank you for your patience.