Statistics Summary for op agent (mean ± std):
==================================================
accuracy: 88.2353 ± 0.0000
total_entries: 136.0000 ± 0.0000
correct_matches: 120.0000 ± 0.0000
average_execution_time: 2.8403 ± 0.0809
metrics_by_complexity:
  simple:
    count: 116.0000 ± 0.0000
    full_matches: 113.0000 ± 0.0000
    exact_code_matches: 109.0000 ± 0.0000
    exact_code_match_rate: 93.9655 ± 0.0000
    pv_exact_matched: 114.0000 ± 0.0000
    pv_exact_mismatched: 2.0000 ± 0.0000
    pv_exact_match_rate: 98.2759 ± 0.0000
    average_pv_match_rate: 99.2816 ± 0.0000
    average_pv_mismatch_rate: 0.7184 ± 0.0000
    timing_matched: 114.0000 ± 0.0000
    timing_mismatched: 2.0000 ± 0.0000
    timing_match_rate: 98.2759 ± 0.0000
    temp_matched: 0.0000 ± 0.0000
    temp_mismatched: 0.0000 ± 0.0000
    temp_match_rate: 0.0000 ± 0.0000
    full_matched: 113.0000 ± 0.0000
    full_mismatched: 3.0000 ± 0.0000
    full_match_rate: 97.4138 ± 0.0000
    accuracy: 97.4138 ± 0.0000
    average_timing_score: 0.9900 ± 0.0000
    average_temp_score: 1.0000 ± 0.0000
    average_full_score: 0.9922 ± 0.0000
    average_best_codebleu:
      average_CodeBLEU: 0.6235 ± 0.0000
      comb_7_CodeBLEU: 0.8465 ± 0.0000
      dataflow_match_score: 0.2845 ± 0.0000
      ngram_match_score: 0.2994 ± 0.0000
      syntax_match_score: 0.9903 ± 0.0000
      weighted_ngram_match_score: 0.2041 ± 0.0000
    average_best_levenshtein: 0.7931 ± 0.0000
    average_best_normalized_levenshtein: 0.0158 ± 0.0000
    average_inference_time: 2.5981 ± 0.0556
  complex:
    count: 20.0000 ± 0.0000
    full_matches: 7.0000 ± 0.0000
    exact_code_matches: 0.0000 ± 0.0000
    exact_code_match_rate: 0.0000 ± 0.0000
    pv_exact_matched: 10.0000 ± 0.0000
    pv_exact_mismatched: 10.0000 ± 0.0000
    pv_exact_match_rate: 50.0000 ± 0.0000
    average_pv_match_rate: 81.5612 ± 0.0000
    average_pv_mismatch_rate: 18.4388 ± 0.0000
    timing_matched: 13.0000 ± 0.0000
    timing_mismatched: 7.0000 ± 0.0000
    timing_match_rate: 65.0000 ± 0.0000
    temp_matched: 8.0000 ± 0.0000
    temp_mismatched: 1.0000 ± 0.0000
    temp_match_rate: 88.8889 ± 0.0000
    full_matched: 7.0000 ± 0.0000
    full_mismatched: 13.0000 ± 0.0000
    full_match_rate: 35.0000 ± 0.0000
    accuracy: 35.0000 ± 0.0000
    average_timing_score: 0.7479 ± 0.0006
    average_temp_score: 0.9807 ± 0.0000
    average_full_score: 0.8036 ± 0.0001
    average_best_codebleu:
      average_CodeBLEU: 0.4508 ± 0.0000
      comb_7_CodeBLEU: 0.5699 ± 0.0000
      dataflow_match_score: 0.5466 ± 0.0000
      ngram_match_score: 0.2289 ± 0.0000
      syntax_match_score: 0.7019 ± 0.0000
      weighted_ngram_match_score: 0.2760 ± 0.0000
    average_best_levenshtein: 156.5500 ± 0.0000
    average_best_normalized_levenshtein: 0.4198 ± 0.0000
    average_inference_time: 4.2450 ± 0.3676
==================================================
