Dear Author:
When I study the paper and compare it to the GitHub code in "generator.py", I found that some functions may be missing from these files, which makes it difficult to clearly understand how self-improvement is achieved. Could you please provide the implementations of "compare_scenario_seq", "compare_scenario_cmb", "filter_inconsistency", etc. in "testbench_parse.py"? Also, could you explain how the scores of "ConsistencyChecker" are computed in "check_consistency"?
Thank you very much for the great help!