When will the dataset be released?

[w-yibo]

thanks for your amazing work!

[Jerry-hyl]

We instruct the annotator model to later step by step, stopping the process once the first error is detected, at which point they provide the corresponding feedback. This strategy effectively decomposes the entire solution, reducing the difficulty of providing comments.

Same. And could you please provide a more detailed explanation of how step-level critique data is generated? The description in your paper seems more like a process of checking for errors step by step and then directly generating the critique data for the entire QA pair.
Thanks!

[iseesaw]

Similarly waiting