Graham Neubig 689d3c9046
Update pre-commit hook versions to most recent versions (#8343)
Co-authored-by: openhands <openhands@all-hands.dev>
2025-05-08 03:59:13 +00:00

29 lines
729 B
Python

import argparse
import json
def main():
parser = argparse.ArgumentParser(description="Get agent's gaia score")
parser.add_argument('--file', type=str, help="Path to the agent's output.jsonl")
args = parser.parse_args()
this_log = args.file
outs = []
with open(this_log, 'r') as f:
lines = f.readlines()
for line in lines:
outs.append(json.loads(line))
print(f'Reading {this_log}')
print(f'Metadata:\n {outs[0]["metadata"]}')
total = 0
success = 0
for out in outs:
total += 1
if out['test_result']['score']:
success += 1
print(f'Success rate: {success}/{total} = {success / total}')
if __name__ == '__main__':
main()