tier1_coupling_results.json

{
  "experiment": "Coupling Constraint Validation",
  "model": "llama3.1:8b",
  "n_queries_per_wer": 10,
  "wer_levels_tested": [
    0,
    1,
    2,
    3,
    5,
    8,
    10,
    15,
    20
  ],
  "results": {
    "wer_0": {
      "wer_pct": 0,
      "mean_quality": 0.9578,
      "std_quality": 0.1268,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        0.5775,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": 0.0,
      "degradation_pct": 0.0
    },
    "wer_1": {
      "wer_pct": 1,
      "mean_quality": 1.0,
      "std_quality": 0.0,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": -0.0422,
      "degradation_pct": -4.41
    },
    "wer_2": {
      "wer_pct": 2,
      "mean_quality": 0.9817,
      "std_quality": 0.0548,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        0.8175,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": -0.0239,
      "degradation_pct": -2.5
    },
    "wer_3": {
      "wer_pct": 3,
      "mean_quality": 0.9828,
      "std_quality": 0.0517,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        0.8275,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": -0.025,
      "degradation_pct": -2.61
    },
    "wer_5": {
      "wer_pct": 5,
      "mean_quality": 1.0,
      "std_quality": 0.0,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": -0.0422,
      "degradation_pct": -4.41
    },
    "wer_8": {
      "wer_pct": 8,
      "mean_quality": 1.0,
      "std_quality": 0.0,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": -0.0422,
      "degradation_pct": -4.41
    },
    "wer_10": {
      "wer_pct": 10,
      "mean_quality": 1.0,
      "std_quality": 0.0,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": -0.0422,
      "degradation_pct": -4.41
    },
    "wer_15": {
      "wer_pct": 15,
      "mean_quality": 1.0,
      "std_quality": 0.0,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0
      ],
      "degradation_from_clean": -0.0422,
      "degradation_pct": -4.41
    },
    "wer_20": {
      "wer_pct": 20,
      "mean_quality": 0.9248,
      "std_quality": 0.1604,
      "n_queries": 10,
      "quality_scores": [
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        1.0,
        0.7475,
        0.5,
        1.0,
        1.0
      ],
      "degradation_from_clean": 0.033,
      "degradation_pct": 3.45
    }
  },
  "critical_wer_threshold": null,
  "conclusion": "No critical threshold found"
}