in ml/eval/bt.py [0:0]
def main():
args = ScriptArguments()
print("Loading data...")
old_rewards = load_rewards(args.sft_generations_file)
new_rewards = load_rewards(args.kto_generations_file)
# Perform Bradley-Terry comparison
print("Performing Bradley-Terry comparison...")
results, metrics = bradley_terry_comparison(old_rewards, new_rewards)
save_results(results, args.output_file)
print_metrics(metrics)