Open depthfirst opened 9 years ago
Added some nice argument parsing. Now --help will give you some help:
usage: ProblemTester.py [-h] [--rerun_problems] [--max_depth int] model_dir problem_dir_or_file [problem_dir_or_file ...]
positional arguments: model_dir location of machine learning models problem_dir_or_file path to problem directory or file
optional arguments: -h, --help show this help message and exit --rerun_problems, -r rerun problems already run (default: False) --max_depth int, -d int maximum depth to search for the next move (default: 3)
Uses machine learning model with search methods to predict moves toward a solution of black-to-live life-and-death Go problems. Final evaluation is if black lives within double the maximum problem path length, score=1, otherwise 0.
Usage: python ProblemTester model_dir problem_dir_or_file1 [problem_dir_or_file2...]
Finds .sgf files and runs them. A problem_test_results.txt file will be written to model_dir. If it already exists, any problems in it will be skipped, and any new problems appended to it.
This is just a sample main() function and should be checked before using.