Statistics Summary for op agent (mean ± std):
==================================================
accuracy: 48.4127 ± 0.0000
total_entries: 126.0000 ± 0.0000
correct_matches: 61.0000 ± 0.0000
average_execution_time: 1.0182 ± 0.1338
metrics_by_complexity:
  simple:
    count: 120.0000 ± 0.0000
    full_matches: 61.0000 ± 0.0000
    exact_code_matches: 30.0000 ± 0.0000
    exact_code_match_rate: 25.0000 ± 0.0000
    pv_exact_matched: 61.0000 ± 0.0000
    pv_exact_mismatched: 59.0000 ± 0.0000
    pv_exact_match_rate: 50.8333 ± 0.0000
    average_pv_match_rate: 60.9028 ± 0.0000
    average_pv_mismatch_rate: 39.0972 ± 0.0000
    timing_matched: 61.0000 ± 0.0000
    timing_mismatched: 59.0000 ± 0.0000
    timing_match_rate: 50.8333 ± 0.0000
    temp_matched: 0.0000 ± 0.0000
    temp_mismatched: 0.0000 ± 0.0000
    temp_match_rate: 0.0000 ± 0.0000
    full_matched: 61.0000 ± 0.0000
    full_mismatched: 59.0000 ± 0.0000
    full_match_rate: 50.8333 ± 0.0000
    accuracy: 50.8333 ± 0.0000
    average_timing_score: 0.5058 ± 0.0000
    average_temp_score: 0.9083 ± 0.0000
    average_full_score: 0.5884 ± 0.0000
    average_best_codebleu:
      codebleu: 0.4870 ± 0.0000
      ngram_match_score: 0.0823 ± 0.0000
      weighted_ngram_match_score: 0.0724 ± 0.0000
      syntax_match_score: 0.7934 ± 0.0000
      dataflow_match_score: 0.1083 ± 0.0000
    average_best_levenshtein: 68.9528 ± 0.0385
    average_best_normalized_levenshtein: 0.3873 ± 0.0001
    average_inference_time: 1.0020 ± 0.1295
  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: 15.2778 ± 0.0000
    average_pv_mismatch_rate: 84.7222 ± 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: 0.1222 ± 0.0000
    average_best_codebleu:
      codebleu: 0.3264 ± 0.0074
      ngram_match_score: 0.0811 ± 0.0020
      weighted_ngram_match_score: 0.1004 ± 0.0029
      syntax_match_score: 0.4626 ± 0.0100
      dataflow_match_score: 0.4948 ± 0.0148
    average_best_levenshtein: 125.0556 ± 5.2924
    average_best_normalized_levenshtein: 0.6450 ± 0.0120
    average_inference_time: 1.3416 ± 0.2209
==================================================
