Evaluation script for mind2web?

I wonder if there is any local evaluation script that includes data preprocessing and parsing of the outputs?

I was trying to create a similar one, but was unable to reproduce the results reported here.

Start by checking the repository for any existing Mind2Web evaluation entry points, data preprocessing, or output-parsing utilities. Compare those paths with the reported Mind2Web results and determine what is missing. Done means a local evaluation script can reproduce the reported results from the available data and model outputs.

Written by the indexing model from the issue text.