In the paper, Table A1, you give out the results on multiple benchmarks. Can you release the detailed codes for acquiring these results?