processed/summary.json

{
  "total_rows": 3010,
  "splits": {
    "train": 2708,
    "valid": 151,
    "test": 151
  },
  "lang_pairs": {
    "en->mas": 1505,
    "mas->en": 1505
  },
  "top_source_names": {
    "synthetic": 2646,
    "cultural_manual": 186,
    "extended_cultural": 148,
    "nkatini_oyete": 30
  },
  "top_domains": {
    "environment": 2046,
    "daily_life": 670,
    "greetings": 44,
    "proverbs": 40,
    "philosophy": 34,
    "livestock": 34,
    "culture": 26,
    "ceremony": 20,
    "education": 20,
    "health": 20
  }
}