Statistics Summary for op agent (mean ± std):
==================================================
accuracy: 32.3529 ± 0.0000
total_entries: 136.0000 ± 0.0000
correct_matches: 44.0000 ± 0.0000
average_execution_time: 1.3472 ± 0.0359
metrics_by_complexity:
  simple:
    count: 116.0000 ± 0.0000
    full_matches: 44.0000 ± 0.0000
    exact_code_matches: 22.0000 ± 0.0000
    exact_code_match_rate: 18.9655 ± 0.0000
    pv_exact_matched: 45.0000 ± 0.0000
    pv_exact_mismatched: 71.0000 ± 0.0000
    pv_exact_match_rate: 38.7931 ± 0.0000
    average_pv_match_rate: 41.6667 ± 0.0000
    average_pv_mismatch_rate: 58.3333 ± 0.0000
    timing_matched: 44.0000 ± 0.0000
    timing_mismatched: 72.0000 ± 0.0000
    timing_match_rate: 37.9310 ± 0.0000
    temp_matched: 0.0000 ± 0.0000
    temp_mismatched: 1.0000 ± 0.0000
    temp_match_rate: 0.0000 ± 0.0000
    full_matched: 44.0000 ± 0.0000
    full_mismatched: 72.0000 ± 0.0000
    full_match_rate: 37.9310 ± 0.0000
    accuracy: 37.9310 ± 0.0000
    average_timing_score: 0.4019 ± 0.0021
    average_temp_score: 0.9914 ± 0.0000
    average_full_score: 0.4126 ± 0.0004
    average_best_codebleu:
      average_CodeBLEU: 0.4354 ± 0.0000
      comb_7_CodeBLEU: 0.6607 ± 0.0000
      dataflow_match_score: 0.2134 ± 0.0000
      ngram_match_score: 0.0643 ± 0.0000
      syntax_match_score: 0.6238 ± 0.0000
      weighted_ngram_match_score: 0.0555 ± 0.0000
    average_best_levenshtein: 315.5259 ± 0.0000
    average_best_normalized_levenshtein: 0.4626 ± 0.0000
    average_inference_time: 1.2696 ± 0.0417
  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: 2.5000 ± 0.0000
    average_pv_mismatch_rate: 97.5000 ± 0.0000
    timing_matched: 0.6667 ± 0.5774
    timing_mismatched: 19.3333 ± 0.5774
    timing_match_rate: 3.3333 ± 2.8868
    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: 0.0414 ± 0.0080
    average_temp_score: 0.5500 ± 0.0000
    average_full_score: 0.0283 ± 0.0016
    average_best_codebleu:
      average_CodeBLEU: 0.3330 ± 0.0000
      comb_7_CodeBLEU: 0.4820 ± 0.0000
      dataflow_match_score: 0.4716 ± 0.0000
      ngram_match_score: 0.0465 ± 0.0000
      syntax_match_score: 0.4912 ± 0.0000
      weighted_ngram_match_score: 0.1228 ± 0.0000
    average_best_levenshtein: 415.2000 ± 0.0000
    average_best_normalized_levenshtein: 0.6682 ± 0.0000
    average_inference_time: 1.7974 ± 0.0025
==================================================
