Statistics Summary for op agent (mean ± std):
==================================================
accuracy: 0.7353 ± 1.2736
total_entries: 136.0000 ± 0.0000
correct_matches: 1.0000 ± 1.7321
average_execution_time: 0.8785 ± 0.3973
metrics_by_complexity:
  simple:
    count: 116.0000 ± 0.0000
    full_matches: 1.0000 ± 1.7321
    exact_code_matches: 1.0000 ± 1.7321
    exact_code_match_rate: 0.8621 ± 1.4931
    pv_exact_matched: 1.0000 ± 1.7321
    pv_exact_mismatched: 115.0000 ± 1.7321
    pv_exact_match_rate: 0.8621 ± 1.4931
    average_pv_match_rate: 1.0057 ± 1.7420
    average_pv_mismatch_rate: 98.9943 ± 1.7420
    timing_matched: 1.3333 ± 2.3094
    timing_mismatched: 114.6667 ± 2.3094
    timing_match_rate: 1.1494 ± 1.9909
    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: 115.0000 ± 1.7321
    full_match_rate: 0.8621 ± 1.4931
    accuracy: 0.8621 ± 1.4931
    average_timing_score: 0.0115 ± 0.0199
    average_temp_score: 1.0000 ± 0.0000
    average_full_score: 0.0103 ± 0.0179
    average_best_codebleu:
      average_CodeBLEU: 0.2585 ± 0.0147
      comb_7_CodeBLEU: 0.4123 ± 0.0214
      dataflow_match_score: 0.0259 ± 0.0448
      ngram_match_score: 0.0023 ± 0.0041
      syntax_match_score: 0.0298 ± 0.0515
      weighted_ngram_match_score: 0.0019 ± 0.0033
    average_best_levenshtein: 172.2011 ± 262.5600
    average_best_normalized_levenshtein: 0.9850 ± 0.0260
    average_inference_time: 0.9178 ± 0.4661
  complex:
    count: 20.0000 ± 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: 20.0000 ± 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: 20.0000 ± 0.0000
    timing_match_rate: 0.0000 ± 0.0000
    temp_matched: 0.0000 ± 0.0000
    temp_mismatched: 9.0000 ± 0.0000
    temp_match_rate: 0.0000 ± 0.0000
    full_matched: 0.0000 ± 0.0000
    full_mismatched: 20.0000 ± 0.0000
    full_match_rate: 0.0000 ± 0.0000
    accuracy: 0.0000 ± 0.0000
    average_timing_score: 0.0000 ± 0.0000
    average_temp_score: 0.5500 ± 0.0000
    average_full_score: 0.0000 ± 0.0000
    average_best_codebleu:
      average_CodeBLEU: 0.2500 ± 0.0000
      comb_7_CodeBLEU: 0.4000 ± 0.0000
      dataflow_match_score: 0.0000 ± 0.0000
      ngram_match_score: 0.0000 ± 0.0000
      syntax_match_score: 0.0000 ± 0.0000
      weighted_ngram_match_score: 0.0000 ± 0.0000
    average_best_levenshtein: 256.7500 ± 0.0000
    average_best_normalized_levenshtein: 1.0000 ± 0.0000
    average_inference_time: 0.6504 ± 0.0018
==================================================
