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: 0.7662 ± 0.3395
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: 12.3472 ± 2.2132
    average_pv_mismatch_rate: 87.6528 ± 2.2132
    timing_matched: 31.3333 ± 8.0829
    timing_mismatched: 88.6667 ± 8.0829
    timing_match_rate: 26.1111 ± 6.7358
    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: 0.2611 ± 0.0674
    average_temp_score: 1.0000 ± 0.0000
    average_full_score: 0.1510 ± 0.0312
    average_best_codebleu:
      average_CodeBLEU: 0.2570 ± 0.0121
      comb_7_CodeBLEU: 0.4100 ± 0.0173
      dataflow_match_score: 0.0250 ± 0.0433
      ngram_match_score: 0.0023 ± 0.0039
      syntax_match_score: 0.0239 ± 0.0414
      weighted_ngram_match_score: 0.0018 ± 0.0032
    average_best_levenshtein: 150.8861 ± 225.6910
    average_best_normalized_levenshtein: 0.9856 ± 0.0249
    average_inference_time: 0.7648 ± 0.3373
  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: 0.2322 ± 0.0309
      comb_7_CodeBLEU: 0.3703 ± 0.0515
      dataflow_match_score: 0.0591 ± 0.1023
      ngram_match_score: 0.0006 ± 0.0010
      syntax_match_score: 0.0323 ± 0.0559
      weighted_ngram_match_score: 0.0033 ± 0.0058
    average_best_levenshtein: 181.5000 ± 107.0985
    average_best_normalized_levenshtein: 0.9550 ± 0.0780
    average_inference_time: 0.7928 ± 0.3837
==================================================
