Statistics Summary for op agent (mean ± std):
==================================================
accuracy: 0.7937 ± 1.3746
total_entries: 126.0000 ± 0.0000
correct_matches: 1.0000 ± 1.7321
average_execution_time: 1.2060 ± 1.2138
metrics_by_complexity:
  simple:
    count: 120.0000 ± 0.0000
    full_matches: 1.0000 ± 1.7321
    exact_code_matches: 1.0000 ± 1.7321
    exact_code_match_rate: 0.8333 ± 1.4434
    pv_exact_matched: 1.0000 ± 1.7321
    pv_exact_mismatched: 119.0000 ± 1.7321
    pv_exact_match_rate: 0.8333 ± 1.4434
    average_pv_match_rate: 12.6528 ± 1.6839
    average_pv_mismatch_rate: 87.3472 ± 1.6839
    timing_matched: 32.6667 ± 5.7735
    timing_mismatched: 87.3333 ± 5.7735
    timing_match_rate: 27.2222 ± 4.8113
    temp_matched: 0.0000 ± 0.0000
    temp_mismatched: 0.0000 ± 0.0000
    temp_match_rate: 0.0000 ± 0.0000
    full_matched: 1.0000 ± 1.7321
    full_mismatched: 119.0000 ± 1.7321
    full_match_rate: 0.8333 ± 1.4434
    accuracy: 0.8333 ± 1.4434
    average_timing_score: 0.2722 ± 0.0481
    average_temp_score: 1.0000 ± 0.0000
    average_full_score: 0.1557 ± 0.0231
    average_best_codebleu:
      average_CodeBLEU: 0.2566 ± 0.0115
      comb_7_CodeBLEU: 0.4094 ± 0.0163
      dataflow_match_score: 0.0222 ± 0.0385
      ngram_match_score: 0.0023 ± 0.0039
      syntax_match_score: 0.0225 ± 0.0389
      weighted_ngram_match_score: 0.0018 ± 0.0032
    average_best_levenshtein: 366.2333 ± 598.6834
    average_best_normalized_levenshtein: 0.9900 ± 0.0174
    average_inference_time: 1.1988 ± 1.2015
  complex:
    count: 6.0000 ± 0.0000
    full_matches: 0.0000 ± 0.0000
    exact_code_matches: 0.0000 ± 0.0000
    exact_code_match_rate: 0.0000 ± 0.0000
    pv_exact_matched: 0.0000 ± 0.0000
    pv_exact_mismatched: 6.0000 ± 0.0000
    pv_exact_match_rate: 0.0000 ± 0.0000
    average_pv_match_rate: 0.0000 ± 0.0000
    average_pv_mismatch_rate: 100.0000 ± 0.0000
    timing_matched: 0.0000 ± 0.0000
    timing_mismatched: 6.0000 ± 0.0000
    timing_match_rate: 0.0000 ± 0.0000
    temp_matched: 0.0000 ± 0.0000
    temp_mismatched: 0.0000 ± 0.0000
    temp_match_rate: 0.0000 ± 0.0000
    full_matched: 0.0000 ± 0.0000
    full_mismatched: 6.0000 ± 0.0000
    full_match_rate: 0.0000 ± 0.0000
    accuracy: 0.0000 ± 0.0000
    average_timing_score: 0.0000 ± 0.0000
    average_temp_score: 1.0000 ± 0.0000
    average_full_score: 0.0000 ± 0.0000
    average_best_codebleu:
      average_CodeBLEU: 0.2456 ± 0.0076
      comb_7_CodeBLEU: 0.3913 ± 0.0150
      dataflow_match_score: 0.2436 ± 0.4219
      ngram_match_score: 0.0005 ± 0.0008
      syntax_match_score: 0.0666 ± 0.1154
      weighted_ngram_match_score: 0.0052 ± 0.0091
    average_best_levenshtein: 380.0556 ± 451.0068
    average_best_normalized_levenshtein: 0.9607 ± 0.0681
    average_inference_time: 1.3498 ± 1.4611
==================================================
