processed/quality_report.json

{
  "metadata": {
    "source_file": "data/processed/all_merged.jsonl",
    "output_directory": "data/processed"
  },
  "summary": {
    "total": 13670,
    "gold": 82,
    "silver": 64,
    "bronze": 13524,
    "issues_distribution": {
      "language_mismatch_target": 7512,
      "misaligned_lengths": 3552,
      "language_mismatch_source": 6012
    },
    "tier_files": {
      "gold": "data/processed/data_gold_tier.jsonl",
      "silver": "data/processed/data_silver_tier.jsonl",
      "bronze": "data/processed/data_bronze_tier.jsonl"
    }
  },
  "issues_distribution": {
    "language_mismatch_target": 7512,
    "misaligned_lengths": 3552,
    "language_mismatch_source": 6012
  },
  "recommendations": [
    "\u26a0\ufe0f  Low gold tier ratio (82/13670). Consider native speaker review for high-confidence pairs before training.",
    "\u26a0\ufe0f  High language detection mismatches. Verify source language coding.",
    "\u26a0\ufe0f  Many length misalignments. Check pair alignment in source data.",
    "\u2139\ufe0f  Contains {bronze} bronze-tier pairs for cultural grounding/instruction tuning only."
  ]
}