frontend/public/results.json

{
  "model_table": [
    {
      "rank": 1,
      "provider": "Google",
      "model": "Gemini 2.0 Flash 001",
      "hf_id": null,
      "creation_date": null,
      "size": null,
      "type": "Commercial",
      "license": null,
      "average": 0.72,
      "classification_accuracy": 0.87,
      "language_modeling_chrf": 0.96,
      "translation_bleu": 0.45,
      "translation_chrf": 0.58
    },
    {
      "rank": 2,
      "provider": "Google",
      "model": "Gemini 2.0 Flash Lite 001",
      "hf_id": null,
      "creation_date": null,
      "size": null,
      "type": "Commercial",
      "license": null,
      "average": 0.66,
      "classification_accuracy": 0.73,
      "language_modeling_chrf": 0.97,
      "translation_bleu": 0.4,
      "translation_chrf": 0.54
    },
    {
      "rank": 3,
      "provider": "Google",
      "model": "Gemma 3 27b It",
      "hf_id": "google/gemma-3-27b-it",
      "creation_date": "2025-03-01",
      "size": 27432406640.0,
      "type": "Open",
      "license": "Gemma",
      "average": 0.65,
      "classification_accuracy": 0.72,
      "language_modeling_chrf": 0.96,
      "translation_bleu": 0.37,
      "translation_chrf": 0.54
    },
    {
      "rank": 4,
      "provider": "Meta Llama",
      "model": "Llama 3.1 70b Instruct",
      "hf_id": "meta-llama/Llama-3.1-70B-Instruct",
      "creation_date": "2024-07-16",
      "size": 70553706496.0,
      "type": "Open",
      "license": "Llama3.1",
      "average": 0.62,
      "classification_accuracy": 0.57,
      "language_modeling_chrf": 0.92,
      "translation_bleu": 0.43,
      "translation_chrf": 0.57
    },
    {
      "rank": 5,
      "provider": "Amazon",
      "model": "Nova Micro V1",
      "hf_id": null,
      "creation_date": null,
      "size": null,
      "type": "Commercial",
      "license": null,
      "average": 0.61,
      "classification_accuracy": 0.52,
      "language_modeling_chrf": 0.94,
      "translation_bleu": 0.4,
      "translation_chrf": 0.56
    },
    {
      "rank": 6,
      "provider": "Meta Llama",
      "model": "Llama 3 70b Instruct",
      "hf_id": null,
      "creation_date": null,
      "size": null,
      "type": "Commercial",
      "license": null,
      "average": 0.61,
      "classification_accuracy": 0.8,
      "language_modeling_chrf": 0.95,
      "translation_bleu": 0.25,
      "translation_chrf": 0.43
    },
    {
      "rank": 7,
      "provider": "OpenAI",
      "model": "GPT 4o Mini",
      "hf_id": null,
      "creation_date": null,
      "size": null,
      "type": "Commercial",
      "license": null,
      "average": 0.6,
      "classification_accuracy": 0.52,
      "language_modeling_chrf": 0.95,
      "translation_bleu": 0.39,
      "translation_chrf": 0.55
    },
    {
      "rank": 8,
      "provider": "MistralAI",
      "model": "Mistral Small 24b Instruct 2501",
      "hf_id": "mistralai/Mistral-Small-24B-Instruct-2501",
      "creation_date": "2025-01-28",
      "size": 23572403200.0,
      "type": "Open",
      "license": "Apache 2.0",
      "average": 0.58,
      "classification_accuracy": 0.55,
      "language_modeling_chrf": 0.86,
      "translation_bleu": 0.38,
      "translation_chrf": 0.52
    },
    {
      "rank": 9,
      "provider": "Meta Llama",
      "model": "Llama 3.3 70b Instruct",
      "hf_id": "meta-llama/Llama-3.3-70B-Instruct",
      "creation_date": "2024-11-26",
      "size": 70553706496.0,
      "type": "Open",
      "license": "Llama3.3",
      "average": 0.56,
      "classification_accuracy": 0.5,
      "language_modeling_chrf": 0.94,
      "translation_bleu": 0.31,
      "translation_chrf": 0.48
    },
    {
      "rank": 10,
      "provider": "MistralAI",
      "model": "Mistral Nemo",
      "hf_id": null,
      "creation_date": null,
      "size": null,
      "type": "Commercial",
      "license": null,
      "average": 0.55,
      "classification_accuracy": 0.5,
      "language_modeling_chrf": 0.88,
      "translation_bleu": 0.32,
      "translation_chrf": 0.49
    },
    {
      "rank": 11,
      "provider": "Microsoft",
      "model": "Phi 4 Multimodal Instruct",
      "hf_id": "microsoft/Phi-4-multimodal-instruct",
      "creation_date": "2025-02-24",
      "size": 5574460384.0,
      "type": "Open",
      "license": "Mit",
      "average": 0.52,
      "classification_accuracy": 0.42,
      "language_modeling_chrf": 0.87,
      "translation_bleu": 0.32,
      "translation_chrf": 0.46
    },
    {
      "rank": 12,
      "provider": "Qwen",
      "model": "Qwq 32b",
      "hf_id": "Qwen/QwQ-32B",
      "creation_date": "2025-03-05",
      "size": 32763876352.0,
      "type": "Open",
      "license": "Apache 2.0",
      "average": 0.25,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.48,
      "translation_bleu": 0.21,
      "translation_chrf": 0.3
    }
  ],
  "language_table": [
    {
      "language_name": "English",
      "autonym": "English",
      "speakers": 1636485840,
      "family": "Indo-European",
      "average": 0.47,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.58,
      "language_modeling_chrf": 0.92,
      "translation_bleu": 0.37,
      "translation_chrf": 0.49
    },
    {
      "language_name": "Chinese",
      "autonym": "中文",
      "speakers": 1304678914,
      "family": "Sino-Tibetan",
      "average": 0.46,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.55,
      "language_modeling_chrf": 0.86,
      "translation_bleu": 0.35,
      "translation_chrf": 0.53
    },
    {
      "language_name": "Hindi",
      "autonym": "हिन्दी",
      "speakers": 546882144,
      "family": "Indo-European",
      "average": 0.46,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.53,
      "language_modeling_chrf": 0.95,
      "translation_bleu": 0.32,
      "translation_chrf": 0.49
    },
    {
      "language_name": "Spanish",
      "autonym": "Español",
      "speakers": 493528077,
      "family": "Indo-European",
      "average": 0.45,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.53,
      "language_modeling_chrf": 0.96,
      "translation_bleu": 0.28,
      "translation_chrf": 0.46
    },
    {
      "language_name": "Arabic",
      "autonym": "العربية",
      "speakers": 351664197,
      "family": "Afro-Asiatic",
      "average": 0.42,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.43,
      "language_modeling_chrf": 0.93,
      "translation_bleu": 0.28,
      "translation_chrf": 0.47
    },
    {
      "language_name": "Urdu",
      "autonym": "اردو",
      "speakers": 290790290,
      "family": "Indo-European",
      "average": 0.42,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.5,
      "language_modeling_chrf": 0.93,
      "translation_bleu": 0.24,
      "translation_chrf": 0.42
    },
    {
      "language_name": "French",
      "autonym": "Français",
      "speakers": 278611507,
      "family": "Indo-European",
      "average": 0.46,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.53,
      "language_modeling_chrf": 0.98,
      "translation_bleu": 0.32,
      "translation_chrf": 0.49
    },
    {
      "language_name": "Bangla",
      "autonym": "বাংলা",
      "speakers": 267193288,
      "family": "Indo-European",
      "average": 0.38,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.4,
      "language_modeling_chrf": 0.9,
      "translation_bleu": 0.2,
      "translation_chrf": 0.41
    },
    {
      "language_name": "Portuguese",
      "autonym": "Português",
      "speakers": 237496885,
      "family": "Indo-European",
      "average": 0.46,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.5,
      "language_modeling_chrf": 0.97,
      "translation_bleu": 0.31,
      "translation_chrf": 0.49
    },
    {
      "language_name": "Punjabi",
      "autonym": "ਪੰਜਾਬੀ",
      "speakers": 203571210,
      "family": "Indo-European",
      "average": 0.42,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.4,
      "language_modeling_chrf": 0.87,
      "translation_bleu": 0.34,
      "translation_chrf": 0.49
    },
    {
      "language_name": "Russian",
      "autonym": "Русский",
      "speakers": 195841151,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Swahili",
      "autonym": "Kiswahili",
      "speakers": 171610296,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Indonesian",
      "autonym": "Bahasa Indonesia",
      "speakers": 171207687,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "German",
      "autonym": "Deutsch",
      "speakers": 136350226,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Japanese",
      "autonym": "日本語",
      "speakers": 119729026,
      "family": "Japonic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Telugu",
      "autonym": "తెలుగు",
      "speakers": 95478480,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Panjabi",
      "autonym": "لہندا پنجابی",
      "speakers": 93433552,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Marathi",
      "autonym": "मराठी",
      "speakers": 92826300,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Javanese",
      "autonym": "Jawa",
      "speakers": 91180665,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Vietnamese",
      "autonym": "Tiếng Việt",
      "speakers": 86222962,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tamil",
      "autonym": "தமிழ்",
      "speakers": 85616159,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Persian",
      "autonym": "فارسی",
      "speakers": 84710459,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Wu Chinese",
      "autonym": "Wu Chinese",
      "speakers": 83641200,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Turkish",
      "autonym": "Türkçe",
      "speakers": 80360704,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Cantonese",
      "autonym": "粵語",
      "speakers": 79654759,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Korean",
      "autonym": "한국어",
      "speakers": 78357046,
      "family": "Koreanic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Italian",
      "autonym": "Italiano",
      "speakers": 70247060,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Filipino",
      "autonym": "Filipino",
      "speakers": 67471096,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Egyptian Arabic",
      "autonym": "Egyptian Arabic",
      "speakers": 66639360,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gujarati",
      "autonym": "ગુજરાતી",
      "speakers": 61721799,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Thai",
      "autonym": "ไทย",
      "speakers": 55181920,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pashto",
      "autonym": "پښتو",
      "speakers": 53542641,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kannada",
      "autonym": "ಕನ್ನಡ",
      "speakers": 49065330,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nigerian Pidgin",
      "autonym": "Naijíriá Píjin",
      "speakers": 44945880,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Malayalam",
      "autonym": "മലയാളം",
      "speakers": 43257484,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Odia",
      "autonym": "ଓଡ଼ିଆ",
      "speakers": 42434880,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Polish",
      "autonym": "Polski",
      "speakers": 41077399,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Xiang Chinese",
      "autonym": "Xiang Chinese",
      "speakers": 40426580,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hausa",
      "autonym": "Hausa",
      "speakers": 40411882,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sindhi",
      "autonym": "سنڌي",
      "speakers": 40329510,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "North Levantine Arabic",
      "autonym": "العامية",
      "speakers": 39031474,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Malay",
      "autonym": "Bahasa Malaysia",
      "speakers": 38097307,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Burmese",
      "autonym": "မြန်မာ",
      "speakers": 36559231,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Amharic",
      "autonym": "አማርኛ",
      "speakers": 35728475,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Algerian Arabic",
      "autonym": "Algerian Arabic",
      "speakers": 35667507,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Oromo",
      "autonym": "Oromoo",
      "speakers": 34897121,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bhojpuri",
      "autonym": "भोजपुरी",
      "speakers": 32934797,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Uzbek",
      "autonym": "O‘Zbek",
      "speakers": 32792780,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Azerbaijani",
      "autonym": "Azərbaycan",
      "speakers": 32446682,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hakka Chinese",
      "autonym": "Hakka Chinese",
      "speakers": 32062460,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sundanese",
      "autonym": "Basa Sunda",
      "speakers": 32043120,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dutch",
      "autonym": "Nederlands",
      "speakers": 31765645,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Moroccan Arabic",
      "autonym": "Moroccan Arabic",
      "speakers": 30938679,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ukrainian",
      "autonym": "Українська",
      "speakers": 29348975,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yoruba",
      "autonym": "Èdè Yorùbá",
      "speakers": 28685568,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Saraiki",
      "autonym": "سرائیکی",
      "speakers": 28020120,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Igbo",
      "autonym": "Igbo",
      "speakers": 27823640,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Min Nan Chinese",
      "autonym": "Min Nan Chinese",
      "speakers": 26486380,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Cebuano",
      "autonym": "Cebuano",
      "speakers": 26203440,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Awadhi",
      "autonym": "Awadhi",
      "speakers": 25862924,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Malagasy",
      "autonym": "Malagasy",
      "speakers": 24260130,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gan Chinese",
      "autonym": "Gan Chinese",
      "speakers": 23698340,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Romanian",
      "autonym": "Română",
      "speakers": 22187408,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bavarian",
      "autonym": "Bavarian",
      "speakers": 22043627,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nepali",
      "autonym": "नेपाली",
      "speakers": 20903374,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Maithili",
      "autonym": "मैथिली",
      "speakers": 19249149,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Assamese",
      "autonym": "অসমীয়া",
      "speakers": 17239170,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nyanja",
      "autonym": "Nyanja",
      "speakers": 17026781,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Somali",
      "autonym": "Soomaali",
      "speakers": 16911645,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Madurese",
      "autonym": "Madurese",
      "speakers": 16822638,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northeastern Thai",
      "autonym": "Northeastern Thai",
      "speakers": 16554576,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rangpuri",
      "autonym": "Rangpuri",
      "speakers": 16274502,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Magahi",
      "autonym": "Magahi",
      "speakers": 15913080,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Haryanvi",
      "autonym": "हरियाणवी",
      "speakers": 15913080,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Marwari",
      "autonym": "Marwari",
      "speakers": 15913080,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Serbian",
      "autonym": "Српски",
      "speakers": 15602410,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sinhala",
      "autonym": "සිංහල",
      "speakers": 15564656,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Khmer",
      "autonym": "ខ្មែរ",
      "speakers": 15065030,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chhattisgarhi",
      "autonym": "Chhattisgarhi",
      "speakers": 14586990,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nigerian Fulfulde",
      "autonym": "Nigerian Fulfulde",
      "speakers": 14339876,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Zulu",
      "autonym": "Isizulu",
      "speakers": 13973830,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kazakh",
      "autonym": "Қазақ Тілі",
      "speakers": 13637392,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Deccan",
      "autonym": "Deccan",
      "speakers": 13128291,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Czech",
      "autonym": "Čeština",
      "speakers": 13045532,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Swedish",
      "autonym": "Svenska",
      "speakers": 12932871,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hungarian",
      "autonym": "Magyar",
      "speakers": 12443430,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Greek",
      "autonym": "Ελληνικά",
      "speakers": 12292242,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Shona",
      "autonym": "Chishona",
      "speakers": 11782503,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Low German",
      "autonym": "Neddersass’Sch",
      "speakers": 11520008,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Akan",
      "autonym": "Akan",
      "speakers": 11442678,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Quechua",
      "autonym": "Runasimi",
      "speakers": 11385851,
      "family": "Quechuan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central Kurdish",
      "autonym": "کوردیی ناوەندی",
      "speakers": 11086549,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kinyarwanda",
      "autonym": "Kinyarwanda",
      "speakers": 11083625,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Wolof",
      "autonym": "Wolof",
      "speakers": 11025494,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tunisian Arabic",
      "autonym": "Tunisian Arabic",
      "speakers": 10549080,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Iloko",
      "autonym": "Ilokano",
      "speakers": 10481376,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Xhosa",
      "autonym": "Isixhosa",
      "speakers": 10182944,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tigrinya",
      "autonym": "ትግርኛ",
      "speakers": 10145911,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Belarusian",
      "autonym": "Беларуская",
      "speakers": 10064517,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Luba-Lulua",
      "autonym": "Luba-Lulua",
      "speakers": 9770880,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tajik",
      "autonym": "Тоҷикӣ",
      "speakers": 9644223,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Umbundu",
      "autonym": "Umbundu",
      "speakers": 9431467,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bambara",
      "autonym": "Bamanakan",
      "speakers": 9385632,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Afrikaans",
      "autonym": "Afrikaans",
      "speakers": 9318845,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hiligaynon",
      "autonym": "Ilonggo",
      "speakers": 9171204,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kikuyu",
      "autonym": "Gikuyu",
      "speakers": 9099743,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Haitian Creole",
      "autonym": "Haitian Creole",
      "speakers": 8964918,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Catalan",
      "autonym": "Català",
      "speakers": 8679139,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hebrew",
      "autonym": "עברית",
      "speakers": 8675480,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sichuan Yi",
      "autonym": "ꆈꌠꉙ",
      "speakers": 8364120,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mossi",
      "autonym": "Mossi",
      "speakers": 8334160,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Baluchi",
      "autonym": "بلۆچی",
      "speakers": 8227887,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sylheti",
      "autonym": "Sylheti",
      "speakers": 8132550,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kimbundu",
      "autonym": "Kimbundu",
      "speakers": 8130575,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Uyghur",
      "autonym": "ئۇيغۇرچە",
      "speakers": 8052967,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Minangkabau",
      "autonym": "Minangkabau",
      "speakers": 8010780,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Swiss German",
      "autonym": "Schwiizertüütsch",
      "speakers": 7956952,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bulgarian",
      "autonym": "Български",
      "speakers": 7878315,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Standard Moroccan Tamazight",
      "autonym": "ⵜⴰⵎⴰⵣⵉⵖⵜ",
      "speakers": 7823574,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Fula",
      "autonym": "Pulaar",
      "speakers": 7788904,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bosnian",
      "autonym": "Bosanski",
      "speakers": 7594468,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rundi",
      "autonym": "Ikirundi",
      "speakers": 7475454,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kanauji",
      "autonym": "Kanauji",
      "speakers": 7426104,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Santali",
      "autonym": "ᱥᱟᱱᱛᱟᱲᱤ",
      "speakers": 7293495,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Danish",
      "autonym": "Dansk",
      "speakers": 7072056,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Turkmen",
      "autonym": "Türkmen Dili",
      "speakers": 6870838,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kurdish",
      "autonym": "Kurdî (Kurmancî)",
      "speakers": 6866757,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Croatian",
      "autonym": "Hrvatski",
      "speakers": 6813164,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Albanian",
      "autonym": "Shqip",
      "speakers": 6791906,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Slovak",
      "autonym": "Slovenčina",
      "speakers": 6680269,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dyula",
      "autonym": "Dyula",
      "speakers": 6667328,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern Thai",
      "autonym": "Northern Thai",
      "speakers": 6621830,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mongolian",
      "autonym": "Монгол",
      "speakers": 6572846,
      "family": "Mongolic-Khitan",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Southern Sotho",
      "autonym": "Sesotho",
      "speakers": 6390567,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Krio",
      "autonym": "Krio",
      "speakers": 6293684,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tachelhit",
      "autonym": "ⵜⴰⵛⵍⵃⵉⵜ",
      "speakers": 6187736,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tswana",
      "autonym": "Tswana",
      "speakers": 6113428,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mewati",
      "autonym": "Mewati",
      "speakers": 6100014,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Luyia",
      "autonym": "Luluhia",
      "speakers": 5888069,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Guarani",
      "autonym": "Avañe’Ẽ",
      "speakers": 5827107,
      "family": "Tupian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Finnish",
      "autonym": "Suomi",
      "speakers": 5736842,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ganda",
      "autonym": "Luganda",
      "speakers": 5622890,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Betawi",
      "autonym": "Betawi",
      "speakers": 5607546,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kashmiri",
      "autonym": "کٲشُر",
      "speakers": 5598085,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Southern Thai",
      "autonym": "Southern Thai",
      "speakers": 5518192,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Norwegian Bokmål",
      "autonym": "Norsk Bokmål",
      "speakers": 5468932,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Norwegian",
      "autonym": "Norsk",
      "speakers": 5467440,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bemba",
      "autonym": "Ichibemba",
      "speakers": 5402246,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Armenian",
      "autonym": "Հայերեն",
      "speakers": 5317273,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern Sotho",
      "autonym": "Northern Sotho",
      "speakers": 5307578,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Luo (Kenya and Tanzania)",
      "autonym": "Dholuo",
      "speakers": 5245734,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tok Pisin",
      "autonym": "Tok Pisin",
      "speakers": 5154217,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lao",
      "autonym": "ລາວ",
      "speakers": 5138706,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sukuma",
      "autonym": "Sukuma",
      "speakers": 5094094,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Konkani",
      "autonym": "कोंकणी",
      "speakers": 4906533,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tsonga",
      "autonym": "Tsonga",
      "speakers": 4880932,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Main-Franconian",
      "autonym": "Main-Franconian",
      "speakers": 4809582,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Balinese",
      "autonym": "Balinese",
      "speakers": 4806468,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ewe",
      "autonym": "Eʋegbe",
      "speakers": 4690857,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Zhuang",
      "autonym": "Vahcuengh",
      "speakers": 4321462,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Buginese",
      "autonym": "Buginese",
      "speakers": 4298211,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mazanderani",
      "autonym": "مازرونی",
      "speakers": 4246165,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Goan Konkani",
      "autonym": "Goan Konkani",
      "speakers": 4243488,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kamba",
      "autonym": "Kikamba",
      "speakers": 4068120,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kalenjin",
      "autonym": "Kalenjin",
      "speakers": 4068120,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Banjar",
      "autonym": "Banjar",
      "speakers": 4010288,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern Hindko",
      "autonym": "Northern Hindko",
      "speakers": 3969517,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Makhuwa",
      "autonym": "Emakhuwa",
      "speakers": 3912766,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gilaki",
      "autonym": "Gilaki",
      "speakers": 3906472,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lombard",
      "autonym": "Lombard",
      "speakers": 3901518,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Zarma",
      "autonym": "Zarmaciine",
      "speakers": 3871308,
      "family": "Songhay",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ndau",
      "autonym": "Ndau",
      "speakers": 3867046,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sidamo",
      "autonym": "Sidamo",
      "speakers": 3783955,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Achinese",
      "autonym": "Achinese",
      "speakers": 3738364,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Shekhawati",
      "autonym": "Shekhawati",
      "speakers": 3713052,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Riffian",
      "autonym": "Tarifit",
      "speakers": 3692411,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Shan",
      "autonym": "တႆး",
      "speakers": 3687984,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lambadi",
      "autonym": "Lambadi",
      "speakers": 3580443,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Garhwali",
      "autonym": "Garhwali",
      "speakers": 3580443,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Georgian",
      "autonym": "ქართული",
      "speakers": 3543646,
      "family": "Kartvelian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Galician",
      "autonym": "Galego",
      "speakers": 3515530,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lingala",
      "autonym": "Lingála",
      "speakers": 3514491,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mandingo",
      "autonym": "Mandingo",
      "speakers": 3511762,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central Atlas Tamazight",
      "autonym": "Tamaziɣt N Laṭlaṣ",
      "speakers": 3485047,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pattani Malay",
      "autonym": "Pattani Malay",
      "speakers": 3448870,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tiv",
      "autonym": "Tiv",
      "speakers": 3424448,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kabyle",
      "autonym": "Taqbaylit",
      "speakers": 3351886,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kyrgyz",
      "autonym": "Кыргызча",
      "speakers": 3338267,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bikol",
      "autonym": "Bikol",
      "speakers": 3275430,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Fon",
      "autonym": "Fon",
      "speakers": 3216150,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gondi",
      "autonym": "Gondi",
      "speakers": 3182616,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Waray",
      "autonym": "Waray",
      "speakers": 3166927,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Southern Kurdish",
      "autonym": "کوردی خوارگ",
      "speakers": 3142162,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Brahui",
      "autonym": "Brahui",
      "speakers": 3035513,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Baoulé",
      "autonym": "Baoulé",
      "speakers": 3022921,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tibetan",
      "autonym": "བོད་སྐད་",
      "speakers": 3006697,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ibibio",
      "autonym": "Ibibio",
      "speakers": 2996392,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Efik",
      "autonym": "Efik",
      "speakers": 2996392,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sango",
      "autonym": "Sängö",
      "speakers": 2935521,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kumaoni",
      "autonym": "Kumaoni",
      "speakers": 2917398,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Aymara",
      "autonym": "Aymara",
      "speakers": 2838620,
      "family": "Aymaran",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nyankole",
      "autonym": "Runyankore",
      "speakers": 2724939,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Jamaican Creole English",
      "autonym": "Jamaican Creole English",
      "speakers": 2668142,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dogri",
      "autonym": "डोगरी",
      "speakers": 2652180,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gusii",
      "autonym": "Ekegusii",
      "speakers": 2622867,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sasak",
      "autonym": "Sasak",
      "speakers": 2590152,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kurukh",
      "autonym": "Kurukh",
      "speakers": 2519571,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pampanga",
      "autonym": "Pampanga",
      "speakers": 2511163,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "West Albay Bikol",
      "autonym": "West Albay Bikol",
      "speakers": 2511163,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lithuanian",
      "autonym": "Lietuvių",
      "speakers": 2488617,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Beja",
      "autonym": "Beja",
      "speakers": 2460326,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Batak Toba",
      "autonym": "Batak Toba",
      "speakers": 2456639,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sadri",
      "autonym": "Sadri",
      "speakers": 2386962,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Waddar",
      "autonym": "Waddar",
      "speakers": 2386962,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Luba-Katanga",
      "autonym": "Tshiluba",
      "speakers": 2340940,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chiga",
      "autonym": "Rukiga",
      "speakers": 2335662,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Soga",
      "autonym": "Olusoga",
      "speakers": 2292409,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Swati",
      "autonym": "Siswati",
      "speakers": 2212379,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hazaragi",
      "autonym": "Hazaragi",
      "speakers": 2161984,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Meru",
      "autonym": "Kĩmĩrũ",
      "speakers": 2141116,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kangri",
      "autonym": "कांगड़ी",
      "speakers": 2121744,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Afar",
      "autonym": "Afar",
      "speakers": 2119663,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Teso",
      "autonym": "Kiteso",
      "speakers": 2082973,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lomwe",
      "autonym": "Lomwe",
      "speakers": 2046678,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Occitan",
      "autonym": "Occitan",
      "speakers": 2040398,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Balochi",
      "autonym": "بلوچی (رخشانی)",
      "speakers": 2037382,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern Luri",
      "autonym": "لۊری شومالی",
      "speakers": 2020512,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Wagdi",
      "autonym": "Wagdi",
      "speakers": 1989135,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tulu",
      "autonym": "Tulu",
      "speakers": 1989135,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Khandesi",
      "autonym": "Khandesi",
      "speakers": 1989135,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tatar",
      "autonym": "Татар",
      "speakers": 1984108,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Slovenian",
      "autonym": "Slovenščina",
      "speakers": 1973181,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Makasar",
      "autonym": "Makasar",
      "speakers": 1949290,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Wolaytta",
      "autonym": "Wolaytta",
      "speakers": 1946034,
      "family": "Ta-Ne-Omotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nyamwezi",
      "autonym": "Nyamwezi",
      "speakers": 1932242,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bodo",
      "autonym": "बर’",
      "speakers": 1856526,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lampung Api",
      "autonym": "Lampung Api",
      "speakers": 1842479,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chuvash",
      "autonym": "Чӑваш",
      "speakers": 1842386,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bashkir",
      "autonym": "Башҡорт Теле",
      "speakers": 1842386,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rohingya",
      "autonym": "𐴌𐴗𐴥𐴝𐴙𐴚𐴒𐴙𐴝",
      "speakers": 1824082,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mende",
      "autonym": "Mende",
      "speakers": 1813083,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tumbuka",
      "autonym": "Tumbuka",
      "speakers": 1780514,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tamashek",
      "autonym": "Tamashek",
      "speakers": 1776965,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "North Ndebele",
      "autonym": "Isindebele",
      "speakers": 1745556,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Masai",
      "autonym": "Maa",
      "speakers": 1734738,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Serer",
      "autonym": "Serer",
      "speakers": 1731004,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nimadi",
      "autonym": "Nimadi",
      "speakers": 1723917,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Timne",
      "autonym": "Timne",
      "speakers": 1722482,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Scots",
      "autonym": "Scots",
      "speakers": 1644028,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lango (Uganda)",
      "autonym": "Lango (Uganda)",
      "speakers": 1643614,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Domari",
      "autonym": "Domari",
      "speakers": 1613543,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Macedonian",
      "autonym": "Македонски",
      "speakers": 1608565,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Acoli",
      "autonym": "Acoli",
      "speakers": 1600361,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central-Eastern Niger Fulfulde",
      "autonym": "Central-Eastern Niger Fulfulde",
      "speakers": 1594068,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bhili",
      "autonym": "Bhili",
      "speakers": 1591308,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pangasinan",
      "autonym": "Pangasinan",
      "speakers": 1528534,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kongo",
      "autonym": "Kongo",
      "speakers": 1526700,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bini",
      "autonym": "Bini",
      "speakers": 1519599,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Maasina Fulfulde",
      "autonym": "Maasina Fulfulde",
      "speakers": 1505612,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Manipuri",
      "autonym": "মৈতৈলোন্",
      "speakers": 1476591,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Abron",
      "autonym": "Abron",
      "speakers": 1467010,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Makonde",
      "autonym": "Chimakonde",
      "speakers": 1463820,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gheg Albanian",
      "autonym": "Gheg Albanian",
      "speakers": 1430250,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Venda",
      "autonym": "Tshivenḓa",
      "speakers": 1391759,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sena",
      "autonym": "Sena",
      "speakers": 1384517,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Venetian",
      "autonym": "Veneto",
      "speakers": 1380829,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Susu",
      "autonym": "Susu",
      "speakers": 1378014,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Norwegian Nynorsk",
      "autonym": "Norsk Nynorsk",
      "speakers": 1366860,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Makhuwa-Meetto",
      "autonym": "Makua",
      "speakers": 1354419,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rajasthani",
      "autonym": "राजस्थानी",
      "speakers": 1326090,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ho",
      "autonym": "Ho",
      "speakers": 1312829,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Maguindanaon",
      "autonym": "Maguindanaon",
      "speakers": 1310172,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mewari",
      "autonym": "Mewari",
      "speakers": 1286307,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bulu",
      "autonym": "Bulu",
      "speakers": 1276270,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Masaaba",
      "autonym": "Masaaba",
      "speakers": 1254337,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mundari",
      "autonym": "Mundari",
      "speakers": 1252287,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Morisyen",
      "autonym": "Kreol Morisien",
      "speakers": 1241433,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Irish",
      "autonym": "Gaeilge",
      "speakers": 1237487,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Fur",
      "autonym": "Fur",
      "speakers": 1230163,
      "family": "Furan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rejang",
      "autonym": "Rejang",
      "speakers": 1228320,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bhilali",
      "autonym": "Bhilali",
      "speakers": 1220003,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tausug",
      "autonym": "Tausug",
      "speakers": 1200991,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kʼicheʼ",
      "autonym": "KʼIcheʼ",
      "speakers": 1200731,
      "family": "Mayan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bakhtiari",
      "autonym": "Bakhtiari",
      "speakers": 1188926,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kpelle",
      "autonym": "Kpɛlɛɛ",
      "speakers": 1186303,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Cebaara Senoufo",
      "autonym": "Cebaara Senoufo",
      "speakers": 1181687,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern Khmer",
      "autonym": "Northern Khmer",
      "speakers": 1172616,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "West Flemish",
      "autonym": "West Flemish",
      "speakers": 1172070,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Soninke",
      "autonym": "Soninke",
      "speakers": 1153651,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Zaza",
      "autonym": "Zaza",
      "speakers": 1148245,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Latvian",
      "autonym": "Latviešu",
      "speakers": 1147550,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dan",
      "autonym": "Dan",
      "speakers": 1099244,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gorontalo",
      "autonym": "Gorontalo",
      "speakers": 1094807,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tigre",
      "autonym": "Tigre",
      "speakers": 1094616,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Basque",
      "autonym": "Euskara",
      "speakers": 1088519,
      "family": null,
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hadothi",
      "autonym": "Hadothi",
      "speakers": 1087394,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kabardian",
      "autonym": "Kabardian",
      "speakers": 1070873,
      "family": "Abkhaz-Adyge",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Khasi",
      "autonym": "Khasi",
      "speakers": 1060872,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sardinian",
      "autonym": "Sardu",
      "speakers": 1060846,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lozi",
      "autonym": "Lozi",
      "speakers": 1045596,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Frafra",
      "autonym": "Frafra",
      "speakers": 1026907,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Najdi Arabic",
      "autonym": "Najdi Arabic",
      "speakers": 1025205,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ronga",
      "autonym": "Ronga",
      "speakers": 1023339,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Southern Luri",
      "autonym": "Southern Luri",
      "speakers": 1019080,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Newari",
      "autonym": "Newari",
      "speakers": 1000821,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Talysh",
      "autonym": "Talysh",
      "speakers": 1000168,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yiddish",
      "autonym": "ייִדיש",
      "speakers": 997214,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bena",
      "autonym": "Hibena",
      "speakers": 995398,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Shambala",
      "autonym": "Kishambaa",
      "speakers": 995398,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kachhi",
      "autonym": "Kachhi",
      "speakers": 994568,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ngaju",
      "autonym": "Ngaju",
      "speakers": 987996,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kita Maninkakan",
      "autonym": "Kita Maninkakan",
      "speakers": 977670,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Jumli",
      "autonym": "Jumli",
      "speakers": 970493,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central Okinawan",
      "autonym": "Central Okinawan",
      "speakers": 966404,
      "family": "Japonic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mon",
      "autonym": "Mon",
      "speakers": 966114,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chimborazo Highland Quichua",
      "autonym": "Chimborazo Highland Quichua",
      "speakers": 963579,
      "family": "Quechuan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kachin",
      "autonym": "Kachin",
      "speakers": 962032,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Southern Hindko",
      "autonym": "Southern Hindko",
      "speakers": 957354,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Limburgish",
      "autonym": "Limburgish",
      "speakers": 950422,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Manyika",
      "autonym": "Manyika",
      "speakers": 945510,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chechen",
      "autonym": "Нохчийн",
      "speakers": 935365,
      "family": "Nakh-Daghestanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kuanyama",
      "autonym": "Kuanyama",
      "speakers": 920524,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "South Ndebele",
      "autonym": "South Ndebele",
      "speakers": 903418,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Welsh",
      "autonym": "Cymraeg",
      "speakers": 884910,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Adangme",
      "autonym": "Adangme",
      "speakers": 880206,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Estonian",
      "autonym": "Eesti",
      "speakers": 878449,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Parsi-Dari",
      "autonym": "Parsi-Dari",
      "speakers": 864342,
      "family": "Bookkeeping",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yucateco",
      "autonym": "Yucateco",
      "speakers": 861955,
      "family": "Mayan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ewondo",
      "autonym": "Ewondo",
      "speakers": 860095,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Komering",
      "autonym": "Komering",
      "speakers": 854483,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tooro",
      "autonym": "Tooro",
      "speakers": 821807,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Garo",
      "autonym": "Garo",
      "speakers": 821563,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ga",
      "autonym": "Gã",
      "speakers": 821526,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mbunga",
      "autonym": "Mbunga",
      "speakers": 819739,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tetum",
      "autonym": "Tetum",
      "speakers": 816395,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Iban",
      "autonym": "Iban",
      "speakers": 816302,
      "family": "Bookkeeping",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Embu",
      "autonym": "Kĩembu",
      "speakers": 802918,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Swabian",
      "autonym": "Swabian",
      "speakers": 801597,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hmong Njua",
      "autonym": "𞄀𞄄𞄰𞄩𞄍𞄜𞄰",
      "speakers": 781687,
      "family": "Hmong-Mien",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kalanga",
      "autonym": "Kalanga",
      "speakers": 770954,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dotyali",
      "autonym": "Dotyali",
      "speakers": 758198,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Picard",
      "autonym": "Picard",
      "speakers": 746330,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Frisian",
      "autonym": "Frysk",
      "speakers": 743057,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chakma",
      "autonym": "𑄌𑄋𑄴𑄟𑄳𑄦",
      "speakers": 729137,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yao",
      "autonym": "Yao",
      "speakers": 722357,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Uab Meto",
      "autonym": "Uab Meto",
      "speakers": 720970,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Capiznon",
      "autonym": "Capiznon",
      "speakers": 720595,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Asu",
      "autonym": "Kipare",
      "speakers": 702634,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tai Dam",
      "autonym": "ꪼꪕꪒꪾ",
      "speakers": 681177,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Walloon",
      "autonym": "Walon",
      "speakers": 679801,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Koyraboro Senni",
      "autonym": "Koyraboro Senni",
      "speakers": 664816,
      "family": "Songhay",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bagheli",
      "autonym": "Bagheli",
      "speakers": 654424,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Asturian",
      "autonym": "Asturianu",
      "speakers": 650205,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Laki",
      "autonym": "Laki",
      "speakers": 645417,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Munda",
      "autonym": "Munda",
      "speakers": 636523,
      "family": "Bookkeeping",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lisu",
      "autonym": "Lisu",
      "speakers": 627309,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "N’Ko",
      "autonym": "ߒߞߏ",
      "speakers": 626370,
      "family": "Artificial Language",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gronings",
      "autonym": "Gronings",
      "speakers": 622094,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mongo",
      "autonym": "Mongo",
      "speakers": 620858,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dangaura Tharu",
      "autonym": "Dangaura Tharu",
      "speakers": 606558,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Neapolitan",
      "autonym": "Neapolitan",
      "speakers": 605306,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nuer",
      "autonym": "Thok Nath",
      "speakers": 591427,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Breton",
      "autonym": "Brezhoneg",
      "speakers": 563140,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pijin",
      "autonym": "Pijin",
      "speakers": 561780,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Réunion Creole French",
      "autonym": "Réunion Creole French",
      "speakers": 559185,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Avaric",
      "autonym": "Avaric",
      "speakers": 552716,
      "family": "Nakh-Daghestanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ndonga",
      "autonym": "Ndonga",
      "speakers": 552315,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ossetic",
      "autonym": "Ирон",
      "speakers": 541444,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Udmurt",
      "autonym": "Udmurt",
      "speakers": 538544,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ligurian",
      "autonym": "Ligure",
      "speakers": 536663,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kabuverdianu",
      "autonym": "Kabuverdianu",
      "speakers": 530762,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rusyn",
      "autonym": "Rusyn",
      "speakers": 527075,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mari",
      "autonym": "Mari",
      "speakers": 524371,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sicilian",
      "autonym": "Sicilianu",
      "speakers": 511702,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Langi",
      "autonym": "Kɨlaangi",
      "speakers": 509409,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Balti",
      "autonym": "Balti",
      "speakers": 502520,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Eastern Huasteca Nahuatl",
      "autonym": "Eastern Huasteca Nahuatl",
      "speakers": 501735,
      "family": "Uto-Aztecan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Huasteca Nahuatl",
      "autonym": "Western Huasteca Nahuatl",
      "speakers": 501735,
      "family": "Uto-Aztecan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Silesian",
      "autonym": "Ślōnski",
      "speakers": 497670,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kara-Kalpak",
      "autonym": "Kara-Kalpak",
      "speakers": 489046,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gujari",
      "autonym": "Gujari",
      "speakers": 467002,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sirmauri",
      "autonym": "Sirmauri",
      "speakers": 464132,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Maltese",
      "autonym": "Malti",
      "speakers": 457267,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bantawa",
      "autonym": "Bantawa",
      "speakers": 454918,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yakut",
      "autonym": "Саха Тыла",
      "speakers": 453510,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Masalit",
      "autonym": "Masalit",
      "speakers": 451060,
      "family": "Maban",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Jju",
      "autonym": "Kaje",
      "speakers": 449459,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Adyghe",
      "autonym": "Adyghe",
      "speakers": 444583,
      "family": "Abkhaz-Adyge",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yemba",
      "autonym": "Yemba",
      "speakers": 443920,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mingrelian",
      "autonym": "Mingrelian",
      "speakers": 439670,
      "family": "Kartvelian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Erzya",
      "autonym": "Эрзянь Кель",
      "speakers": 439338,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Taita",
      "autonym": "Kitaita",
      "speakers": 438929,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central Mazahua",
      "autonym": "Central Mazahua",
      "speakers": 437410,
      "family": "Otomanguean",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Vunjo",
      "autonym": "Kyivunjo",
      "speakers": 433291,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rombo",
      "autonym": "Kihorombo",
      "speakers": 433291,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Machame",
      "autonym": "Kimachame",
      "speakers": 433291,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Khmu",
      "autonym": "Khmu",
      "speakers": 431949,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Fang",
      "autonym": "Fang",
      "speakers": 426451,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kinaray-a",
      "autonym": "Kinaray-A",
      "speakers": 425806,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kuy",
      "autonym": "Kuy",
      "speakers": 421207,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Luxembourgish",
      "autonym": "Lëtzebuergesch",
      "speakers": 421015,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sranan Tongo",
      "autonym": "Sranan Tongo",
      "speakers": 414507,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Jola-Fonyi",
      "autonym": "Joola",
      "speakers": 409146,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Tamang",
      "autonym": "Western Tamang",
      "speakers": 394263,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pontic",
      "autonym": "Pontic",
      "speakers": 392463,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Negeri Sembilan Malay",
      "autonym": "Negeri Sembilan Malay",
      "speakers": 391825,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ngiemboon",
      "autonym": "Shwóŋò Ngiembɔɔn",
      "speakers": 388430,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ghomala",
      "autonym": "Ghomala",
      "speakers": 388430,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Divehi",
      "autonym": "Divehi",
      "speakers": 388044,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Saurashtra",
      "autonym": "Saurashtra",
      "speakers": 384566,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Fiji Hindi",
      "autonym": "Fiji Hindi",
      "speakers": 383749,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nobiin",
      "autonym": "Nobiin",
      "speakers": 378161,
      "family": "Nubian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Balkan Gagauz Turkish",
      "autonym": "Balkan Gagauz Turkish",
      "speakers": 377280,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kerinci",
      "autonym": "Kerinci",
      "speakers": 373836,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Parkari Koli",
      "autonym": "Parkari Koli",
      "speakers": 373602,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dzongkha",
      "autonym": "རྫོང་ཁ",
      "speakers": 370341,
      "family": "Bookkeeping",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pökoot",
      "autonym": "Pökoot",
      "speakers": 369343,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dargwa",
      "autonym": "Dargwa",
      "speakers": 368477,
      "family": "Nakh-Daghestanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Limbu",
      "autonym": "Limbu",
      "speakers": 368085,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Fijian",
      "autonym": "Fijian",
      "speakers": 365030,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rana Tharu",
      "autonym": "Rana Tharu",
      "speakers": 363935,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kom",
      "autonym": "Kom",
      "speakers": 360685,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Icelandic",
      "autonym": "Íslenska",
      "speakers": 350734,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Khowar",
      "autonym": "Khowar",
      "speakers": 350252,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tolaki",
      "autonym": "Tolaki",
      "speakers": 347134,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Naxi",
      "autonym": "Naxi",
      "speakers": 334565,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Eastern Magar",
      "autonym": "Eastern Magar",
      "speakers": 333607,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bamun",
      "autonym": "Bamun",
      "speakers": 332940,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Basaa",
      "autonym": "Ɓàsàa",
      "speakers": 332940,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Koyra Chiini",
      "autonym": "Koyra Ciini",
      "speakers": 332408,
      "family": "Songhay",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Indus Kohistani",
      "autonym": "Indus Kohistani",
      "speakers": 326901,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gayo",
      "autonym": "Gayo",
      "speakers": 320431,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ngazidja Comorian",
      "autonym": "Ngazidja Comorian",
      "speakers": 313124,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Buriat",
      "autonym": "Buriat",
      "speakers": 311788,
      "family": "Mongolic-Khitan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rinconada Bikol",
      "autonym": "Rinconada Bikol",
      "speakers": 305707,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Medumba",
      "autonym": "Medumba",
      "speakers": 305195,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ao Naga",
      "autonym": "Ao Naga",
      "speakers": 305001,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Badaga",
      "autonym": "Badaga",
      "speakers": 305001,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kochila Tharu",
      "autonym": "Kochila Tharu",
      "speakers": 303279,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Moksha",
      "autonym": "Мокшень Кяль",
      "speakers": 297616,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tae'",
      "autonym": "Tae'",
      "speakers": 293729,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nzima",
      "autonym": "Nzima",
      "speakers": 293402,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nama",
      "autonym": "Khoekhoegowab",
      "speakers": 289308,
      "family": "Khoe-Kwadi",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ndzwani Comorian",
      "autonym": "Ndzwani Comorian",
      "speakers": 287736,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kumyk",
      "autonym": "Kumyk",
      "speakers": 283444,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mundang",
      "autonym": "Mundaŋ",
      "speakers": 277450,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mapuche",
      "autonym": "Mapudungun",
      "speakers": 272802,
      "family": "Araucanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Cham",
      "autonym": "Western Cham",
      "speakers": 270832,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bislama",
      "autonym": "Bislama",
      "speakers": 268500,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lü",
      "autonym": "Lü",
      "speakers": 264864,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tai Nüa",
      "autonym": "Tai Nüa",
      "speakers": 264864,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Wadiyara Koli",
      "autonym": "Wadiyara Koli",
      "speakers": 256851,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kachi Koli",
      "autonym": "Kachi Koli",
      "speakers": 256851,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lezghian",
      "autonym": "Lezghian",
      "speakers": 255100,
      "family": "Nakh-Daghestanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Komi",
      "autonym": "Komi",
      "speakers": 255100,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Samoan",
      "autonym": "Samoan",
      "speakers": 252717,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Magar",
      "autonym": "Western Magar",
      "speakers": 251722,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Samburu",
      "autonym": "Kisampur",
      "speakers": 246228,
      "family": "Nilotic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Crimean Tatar",
      "autonym": "Crimean Tatar",
      "speakers": 245968,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mandar",
      "autonym": "Mandar",
      "speakers": 245664,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sangir",
      "autonym": "Sangir",
      "speakers": 245664,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Extremaduran",
      "autonym": "Extremaduran",
      "speakers": 245077,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central Huasteca Nahuatl",
      "autonym": "Central Huasteca Nahuatl",
      "speakers": 244435,
      "family": "Uto-Aztecan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Zeelandic",
      "autonym": "Zeelandic",
      "speakers": 241926,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Karachay-Balkar",
      "autonym": "Karachay-Balkar",
      "speakers": 240927,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Colognian",
      "autonym": "Kölsch",
      "speakers": 240479,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Herero",
      "autonym": "Herero",
      "speakers": 239336,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Saafi-Saafi",
      "autonym": "Saafi-Saafi",
      "speakers": 236046,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Zaghawa",
      "autonym": "Zaghawa",
      "speakers": 232364,
      "family": "Saharan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ingush",
      "autonym": "Ingush",
      "speakers": 226755,
      "family": "Nakh-Daghestanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Saho",
      "autonym": "Saho",
      "speakers": 218923,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tomo Kan Dogon",
      "autonym": "Tomo Kan Dogon",
      "speakers": 215087,
      "family": "Dogon",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Papiamento",
      "autonym": "Papiamentu",
      "speakers": 211640,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Syriac",
      "autonym": "ܣܘܪܝܝܐ",
      "speakers": 210659,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nyasa Tonga",
      "autonym": "Nyasa Tonga",
      "speakers": 207727,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mafa",
      "autonym": "Mafa",
      "speakers": 205313,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Punu",
      "autonym": "Punu",
      "speakers": 200782,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tyap",
      "autonym": "Katab",
      "speakers": 199046,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Xaasongaxango",
      "autonym": "Xaasongaxango",
      "speakers": 195534,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tuvinian",
      "autonym": "Tuvinian",
      "speakers": 184239,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central Dusun",
      "autonym": "Central Dusun",
      "speakers": 182852,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Riang (India)",
      "autonym": "Riang (India)",
      "speakers": 172392,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Comorian",
      "autonym": "Comorian",
      "speakers": 170720,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bomu",
      "autonym": "Bomu",
      "speakers": 168159,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Latgalian",
      "autonym": "Latgalian",
      "speakers": 167429,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Navajo",
      "autonym": "Diné Bizaad",
      "speakers": 166320,
      "family": "Athabaskan-Eyak-Tlingit",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Jenaama Bozo",
      "autonym": "Jenaama Bozo",
      "speakers": 166204,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Corsican",
      "autonym": "Corsu",
      "speakers": 162836,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bafut",
      "autonym": "Bafut",
      "speakers": 158146,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sherpa",
      "autonym": "Sherpa",
      "speakers": 157705,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kirmanjki",
      "autonym": "Kirmanjki",
      "speakers": 155833,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hiri Motu",
      "autonym": "Hiri Motu",
      "speakers": 152449,
      "family": "Pidgin",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kako",
      "autonym": "Kakɔ",
      "speakers": 149823,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Akoose",
      "autonym": "Akoose",
      "speakers": 149823,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Selayar",
      "autonym": "Selayar",
      "speakers": 144194,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Māori",
      "autonym": "Māori",
      "speakers": 137913,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rajbanshi",
      "autonym": "Rajbanshi",
      "speakers": 133443,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Duala",
      "autonym": "Duálá",
      "speakers": 133176,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Wayuu",
      "autonym": "Wayuu",
      "speakers": 132529,
      "family": "Arawakan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Vai",
      "autonym": "ꕙꔤ",
      "speakers": 131906,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Eastern Tamang",
      "autonym": "Eastern Tamang",
      "speakers": 130410,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Metaʼ",
      "autonym": "Metaʼ",
      "speakers": 130401,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pennsylvania German",
      "autonym": "Pennsylvania German",
      "speakers": 129729,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rwa",
      "autonym": "Kiruwa",
      "speakers": 128816,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Torwali",
      "autonym": "توروالی",
      "speakers": 123756,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mandjak",
      "autonym": "Mandjak",
      "speakers": 121170,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tshangla",
      "autonym": "Tshangla",
      "speakers": 117348,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sangu",
      "autonym": "Ishisangu",
      "speakers": 117106,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Karelian",
      "autonym": "Karelian",
      "speakers": 116212,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ladino",
      "autonym": "Ladino",
      "speakers": 112781,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ifè",
      "autonym": "Ifè",
      "speakers": 111910,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gagauz",
      "autonym": "Gagauz",
      "speakers": 111028,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lak",
      "autonym": "Lak",
      "speakers": 110543,
      "family": "Nakh-Daghestanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sassarese Sardinian",
      "autonym": "Sassarese Sardinian",
      "speakers": 106085,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tongan",
      "autonym": "Lea Fakatonga",
      "speakers": 100790,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Balanta-Ganja",
      "autonym": "Balanta-Ganja",
      "speakers": 95992,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ngomba",
      "autonym": "Ndaꞌa",
      "speakers": 94333,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Seselwa Creole French",
      "autonym": "Seselwa Creole French",
      "speakers": 94061,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Abkhazian",
      "autonym": "Аԥсшәа",
      "speakers": 91953,
      "family": "Abkhaz-Adyge",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tahitian",
      "autonym": "Tahitian",
      "speakers": 91488,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Inuktitut",
      "autonym": "Inuktitut",
      "speakers": 90466,
      "family": "Eskimo-Aleut",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Plautdietsch",
      "autonym": "Plautdietsch",
      "speakers": 90466,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bishnupriya",
      "autonym": "Bishnupriya",
      "speakers": 90174,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bafia",
      "autonym": "Rikpa",
      "speakers": 88784,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gurung",
      "autonym": "Gurung",
      "speakers": 87951,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Eastern Cham",
      "autonym": "Eastern Cham",
      "speakers": 87862,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Lawa",
      "autonym": "Western Lawa",
      "speakers": 87751,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mankanya",
      "autonym": "Mankanya",
      "speakers": 83151,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lepcha",
      "autonym": "Lepcha",
      "speakers": 79743,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Blin",
      "autonym": "Blin",
      "speakers": 79056,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bateri",
      "autonym": "Bateri",
      "speakers": 78843,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kathoriya Tharu",
      "autonym": "Kathoriya Tharu",
      "speakers": 72787,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Scottish Gaelic",
      "autonym": "Gàidhlig",
      "speakers": 72337,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Faroese",
      "autonym": "Føroyskt",
      "speakers": 71351,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Võro",
      "autonym": "Võro",
      "speakers": 70031,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kenyang",
      "autonym": "Kɛnyaŋ",
      "speakers": 69362,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gilbertese",
      "autonym": "Gilbertese",
      "speakers": 67078,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bube",
      "autonym": "Bube",
      "speakers": 66058,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mentawai",
      "autonym": "Mentawai",
      "speakers": 64086,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Arpitan",
      "autonym": "Arpitan",
      "speakers": 63777,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Komi-Permyak",
      "autonym": "Komi-Permyak",
      "speakers": 63775,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Koro",
      "autonym": "Koro",
      "speakers": 63207,
      "family": "Mande",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Marshallese",
      "autonym": "Marshallese",
      "speakers": 56879,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tornedalen Finnish",
      "autonym": "Tornedalen Finnish",
      "speakers": 56114,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kalaallisut",
      "autonym": "Kalaallisut",
      "speakers": 55440,
      "family": "Eskimo-Aleut",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Braj",
      "autonym": "Braj",
      "speakers": 54370,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Guianese Creole French",
      "autonym": "Guianese Creole French",
      "speakers": 51872,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern Sami",
      "autonym": "Davvisámegiella",
      "speakers": 51530,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Anii",
      "autonym": "Anii Kagɩja",
      "speakers": 51507,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kaingang",
      "autonym": "Kanhgág",
      "speakers": 50812,
      "family": "Nuclear-Macro-Je",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kashubian",
      "autonym": "Kashubian",
      "speakers": 49767,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Koro Wachi",
      "autonym": "Koro Wachi",
      "speakers": 46718,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chamorro",
      "autonym": "Chamorro",
      "speakers": 46325,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Atsam",
      "autonym": "Atsam",
      "speakers": 44946,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bushi",
      "autonym": "Bushi",
      "speakers": 44620,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Romansh",
      "autonym": "Rumantsch",
      "speakers": 42020,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Aghem",
      "autonym": "Aghem",
      "speakers": 38843,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kuvi",
      "autonym": "Kuvi",
      "speakers": 38457,
      "family": "Dravidian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Noon",
      "autonym": "Noon",
      "speakers": 37767,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Friulian",
      "autonym": "Furlan",
      "speakers": 37442,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Thulung",
      "autonym": "Thulung",
      "speakers": 36393,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Emilian",
      "autonym": "Emilian",
      "speakers": 31201,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chuukese",
      "autonym": "Chuukese",
      "speakers": 30731,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Western Mari",
      "autonym": "Western Mari",
      "speakers": 29762,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hawaiian",
      "autonym": "ʻŌlelo HawaiʻI",
      "speakers": 29605,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mru",
      "autonym": "Mru",
      "speakers": 29277,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Cajun French",
      "autonym": "Cajun French",
      "speakers": 27942,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nheengatu",
      "autonym": "Nheẽgatu",
      "speakers": 26171,
      "family": "Tupian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Aragonese",
      "autonym": "Aragonés",
      "speakers": 26008,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Cherokee",
      "autonym": "Ꮳꮃꭹ",
      "speakers": 25613,
      "family": "Iroquoian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sinte Romani",
      "autonym": "Sinte Romani",
      "speakers": 24372,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ojibwa",
      "autonym": "Ojibwa",
      "speakers": 23747,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Pohnpeian",
      "autonym": "Pohnpeian",
      "speakers": 23560,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Laz",
      "autonym": "Laz",
      "speakers": 22965,
      "family": "Kartvelian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Muslim Tat",
      "autonym": "Muslim Tat",
      "speakers": 22453,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Central Yupik",
      "autonym": "Central Yupik",
      "speakers": 20956,
      "family": "Eskimo-Aleut",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dakota",
      "autonym": "Dakota",
      "speakers": 20832,
      "family": "Siouan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Southern Altai",
      "autonym": "Southern Altai",
      "speakers": 19841,
      "family": "Turkic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Amo",
      "autonym": "Amo",
      "speakers": 18620,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Guajajára",
      "autonym": "Guajajára",
      "speakers": 17784,
      "family": "Tupian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hanunoo",
      "autonym": "Hanunoo",
      "speakers": 17469,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tsakhur",
      "autonym": "Tsakhur",
      "speakers": 16329,
      "family": "Nakh-Daghestanian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Palauan",
      "autonym": "Palauan",
      "speakers": 16047,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Sanskrit",
      "autonym": "संस्कृत भाषा",
      "speakers": 15913,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Bassari",
      "autonym": "Bassari",
      "speakers": 15264,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Oji-Cree",
      "autonym": "Oji-Cree",
      "speakers": 15078,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Khamti",
      "autonym": "Khamti",
      "speakers": 13527,
      "family": "Tai-Kadai",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Upper Sorbian",
      "autonym": "Hornjoserbšćina",
      "speakers": 12826,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chipewyan",
      "autonym": "Chipewyan",
      "speakers": 12816,
      "family": "Athabaskan-Eyak-Tlingit",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Innu-aimun",
      "autonym": "Innu-Aimun",
      "speakers": 12062,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lower Silesian",
      "autonym": "Lower Silesian",
      "speakers": 11868,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Walser",
      "autonym": "Walser",
      "speakers": 11377,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Choctaw",
      "autonym": "Chahta",
      "speakers": 10977,
      "family": "Muskogean",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tagbanwa",
      "autonym": "Tagbanwa",
      "speakers": 10045,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Xavánte",
      "autonym": "Xavánte",
      "speakers": 9951,
      "family": "Nuclear-Macro-Je",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tuvalu",
      "autonym": "Tuvalu",
      "speakers": 9868,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern Frisian",
      "autonym": "Nordfriisk",
      "speakers": 9619,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Roviana",
      "autonym": "Roviana",
      "speakers": 9591,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Wallisian",
      "autonym": "Wallisian",
      "speakers": 9512,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tavringer Romani",
      "autonym": "Tavringer Romani",
      "speakers": 9488,
      "family": "Speech Register",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Cree",
      "autonym": "Cree",
      "speakers": 9047,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kwasio",
      "autonym": "Kwasio",
      "speakers": 8878,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lakota",
      "autonym": "LakȟólʼIyapi",
      "speakers": 8316,
      "family": "Siouan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kosraean",
      "autonym": "Kosraean",
      "speakers": 7990,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Inupiaq",
      "autonym": "Inupiaq",
      "speakers": 7983,
      "family": "Eskimo-Aleut",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Zoroastrian Dari",
      "autonym": "Zoroastrian Dari",
      "speakers": 7983,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tasawaq",
      "autonym": "Tasawaq Senni",
      "speakers": 7970,
      "family": "Songhay",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Buhid",
      "autonym": "Buhid",
      "speakers": 7970,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mi'kmaw",
      "autonym": "LʼNuiʼSuti",
      "speakers": 7916,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Hassaniyya",
      "autonym": "Hassaniyya",
      "speakers": 7239,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lower Sorbian",
      "autonym": "Dolnoserbšćina",
      "speakers": 6974,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Nauru",
      "autonym": "Nauru",
      "speakers": 6930,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Eastern Lawa",
      "autonym": "Eastern Lawa",
      "speakers": 6898,
      "family": "Austroasiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yapese",
      "autonym": "Yapese",
      "speakers": 6556,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Atikamekw",
      "autonym": "Atikamekw",
      "speakers": 6408,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Piedmontese",
      "autonym": "Piedmontese",
      "speakers": 6178,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kalo Finnish Romani",
      "autonym": "Kalo Finnish Romani",
      "speakers": 5015,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Siksiká",
      "autonym": "Siksiká",
      "speakers": 4900,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "East Futuna",
      "autonym": "East Futuna",
      "speakers": 4756,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Taroko",
      "autonym": "Patas Taroko",
      "speakers": 4721,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Plains Cree",
      "autonym": "Plains Cree",
      "speakers": 4146,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Safaliba",
      "autonym": "Safaliba",
      "speakers": 4108,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Muscogee",
      "autonym": "Mvskoke",
      "speakers": 3992,
      "family": "Muskogean",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Veps",
      "autonym": "Veps",
      "speakers": 3543,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ménik",
      "autonym": "Ménik",
      "speakers": 3305,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Virgin Islands Creole English",
      "autonym": "Virgin Islands Creole English",
      "speakers": 3113,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Turoyo",
      "autonym": "Turoyo",
      "speakers": 3035,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ulithian",
      "autonym": "Ulithian",
      "speakers": 2971,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Rotuman",
      "autonym": "Rotuman",
      "speakers": 2527,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Warlpiri",
      "autonym": "Warlpiri",
      "speakers": 2496,
      "family": "Pama-Nyungan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Yangben",
      "autonym": "Nuasue",
      "speakers": 2303,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Slave",
      "autonym": "Slave",
      "speakers": 2299,
      "family": "Athabaskan-Eyak-Tlingit",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Dogrib",
      "autonym": "Dogrib",
      "speakers": 2111,
      "family": "Athabaskan-Eyak-Tlingit",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Eastern Frisian",
      "autonym": "Eastern Frisian",
      "speakers": 2004,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Cornish",
      "autonym": "Kernewek",
      "speakers": 1973,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Swampy Cree",
      "autonym": "ᓀᐦᐃᓇᐍᐏᐣ",
      "speakers": 1809,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Mohawk",
      "autonym": "KanienʼKéha",
      "speakers": 1772,
      "family": "Iroquoian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Manx",
      "autonym": "Gaelg",
      "speakers": 1719,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lule Sami",
      "autonym": "Julevsámegiella",
      "speakers": 1530,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Naskapi",
      "autonym": "Naskapi",
      "speakers": 1395,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tokelau",
      "autonym": "Tokelau",
      "speakers": 1285,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Niuean",
      "autonym": "Niuean",
      "speakers": 1120,
      "family": "Austronesian",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Saterland Frisian",
      "autonym": "Saterland Frisian",
      "speakers": 962,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Seri",
      "autonym": "Seri",
      "speakers": 901,
      "family": null,
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chilcotin",
      "autonym": "Chilcotin",
      "speakers": 867,
      "family": "Athabaskan-Eyak-Tlingit",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Latin",
      "autonym": "Lingua Latina",
      "speakers": 820,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Halkomelem",
      "autonym": "Halkomelem",
      "speakers": 716,
      "family": "Salishan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Michif",
      "autonym": "Michif",
      "speakers": 678,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Skolt Sami",
      "autonym": "SääʹMǩiõll",
      "speakers": 613,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Inari Sami",
      "autonym": "Anarâškielâ",
      "speakers": 613,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lillooet",
      "autonym": "Lillooet",
      "speakers": 528,
      "family": "Salishan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Okanagan",
      "autonym": "Okanagan",
      "speakers": 490,
      "family": "Salishan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Maliseet-Passamaquoddy",
      "autonym": "Maliseet-Passamaquoddy",
      "speakers": 490,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Kwakʼwala",
      "autonym": "KwakʼWala",
      "speakers": 377,
      "family": "Wakashan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Northern East Cree",
      "autonym": "Northern East Cree",
      "speakers": 377,
      "family": "Algic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Gwichʼin",
      "autonym": "GwichʼIn",
      "speakers": 302,
      "family": "Athabaskan-Eyak-Tlingit",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Esperanto",
      "autonym": "Esperanto",
      "speakers": 301,
      "family": "Artificial Language",
      "average": 0.0,
      "in_benchmark": true,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Southern Sami",
      "autonym": "Åarjelsaemien Gïele",
      "speakers": 296,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Tsakonian",
      "autonym": "Tsakonian",
      "speakers": 202,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ingrian",
      "autonym": "Ingrian",
      "speakers": 142,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Interlingua",
      "autonym": "Interlingua",
      "speakers": 136,
      "family": "Artificial Language",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Araona",
      "autonym": "Araona",
      "speakers": 105,
      "family": "Pano-Tacanan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Prussian",
      "autonym": "Prūsiskan",
      "speakers": 38,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Interlingue",
      "autonym": "Interlingue",
      "speakers": 1,
      "family": "Artificial Language",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Literary Chinese",
      "autonym": "Literary Chinese",
      "speakers": 0,
      "family": "Sino-Tibetan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Ido",
      "autonym": "Ido",
      "speakers": 0,
      "family": "Artificial Language",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Lojban",
      "autonym": "La .Lojban.",
      "speakers": 0,
      "family": "Artificial Language",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Jutish",
      "autonym": "Jutish",
      "speakers": 0,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Votic",
      "autonym": "Votic",
      "speakers": 0,
      "family": "Uralic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Geez",
      "autonym": "Geez",
      "speakers": 0,
      "family": "Afro-Asiatic",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Osage",
      "autonym": "𐓏𐓘𐓻𐓘𐓻𐓟",
      "speakers": 0,
      "family": "Siouan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Romagnol",
      "autonym": "Romagnol",
      "speakers": 0,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Church Slavic",
      "autonym": "Church Slavic",
      "speakers": 0,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Samogitian",
      "autonym": "Samogitian",
      "speakers": 0,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Obolo",
      "autonym": "Obolo",
      "speakers": 0,
      "family": "Atlantic-Congo",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Chickasaw",
      "autonym": "Chikashshanompaʼ",
      "speakers": 0,
      "family": "Muskogean",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Caddo",
      "autonym": "Caddo",
      "speakers": 0,
      "family": "Caddoan",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    },
    {
      "language_name": "Palatine German",
      "autonym": "Palatine German",
      "speakers": 0,
      "family": "Indo-European",
      "average": 0.0,
      "in_benchmark": false,
      "NaN": 0.0,
      "classification_accuracy": 0.0,
      "language_modeling_chrf": 0.0,
      "translation_bleu": 0.0,
      "translation_chrf": 0.0
    }
  ],
  "dataset_table": [
    {
      "name": "FLORES+",
      "author": "Meta",
      "author_url": "https://ai.meta.com",
      "url": "https://huggingface.co/datasets/openlanguagedata/flores_plus",
      "n_languages": 200,
      "tasks": [
        "translation",
        "classification",
        "language_modeling"
      ],
      "parallel": 1.0,
      "base": "FLORES",
      "implemented": 1.0
    },
    {
      "name": "FLEURS",
      "author": "Meta",
      "author_url": "https://ai.meta.com",
      "url": "https://huggingface.co/datasets/google/fleurs",
      "n_languages": 102,
      "tasks": [
        "speech_recognition"
      ],
      "parallel": 1.0,
      "base": "FLORES",
      "implemented": 1.0
    },
    {
      "name": "CommonVoice",
      "author": "Mozilla",
      "author_url": "https://mozilla.ai",
      "url": "https://huggingface.co/datasets/mozilla-foundation/common_voice_17_0",
      "n_languages": 124,
      "tasks": [
        "speech_recognition"
      ],
      "parallel": null,
      "base": null,
      "implemented": null
    },
    {
      "name": "MMMLU",
      "author": "OpenAI",
      "author_url": "https://openai.com",
      "url": "https://huggingface.co/datasets/openai/MMMLU",
      "n_languages": 14,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "MMLU",
      "implemented": null
    },
    {
      "name": "AfriMMLU",
      "author": "Masakhane",
      "author_url": "https://www.masakhane.io",
      "url": "https://huggingface.co/datasets/masakhane/afrimmlu",
      "n_languages": 17,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "MMLU",
      "implemented": null
    },
    {
      "name": "Okapi MMLU",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/jon-tow/okapi_mmlu",
      "n_languages": 16,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "MMLU",
      "implemented": null
    },
    {
      "name": "Global MMLU",
      "author": "Cohere",
      "author_url": "https://cohere.com",
      "url": "https://huggingface.co/datasets/CohereForAI/Global-MMLU",
      "n_languages": 42,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "MMLU",
      "implemented": null
    },
    {
      "name": "MGSM",
      "author": "Google",
      "author_url": "https://google.com",
      "url": "https://huggingface.co/datasets/juletxara/mgsm",
      "n_languages": 10,
      "tasks": [
        "math"
      ],
      "parallel": 1.0,
      "base": "MGSM",
      "implemented": null
    },
    {
      "name": "AfriMGSM",
      "author": "Masakhane",
      "author_url": "https://www.masakhane.io",
      "url": "https://huggingface.co/datasets/masakhane/afrimgsm",
      "n_languages": 18,
      "tasks": [
        "math"
      ],
      "parallel": 1.0,
      "base": "MGSM",
      "implemented": null
    },
    {
      "name": "Okapi ARC Challenge",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/jon-tow/okapi_arc_challenge",
      "n_languages": 31,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "AI2 ARC",
      "implemented": null
    },
    {
      "name": "Uhuru ARC Easy",
      "author": "Masakhane",
      "author_url": "https://www.masakhane.io",
      "url": "https://huggingface.co/datasets/masakhane/uhura-arc-easy",
      "n_languages": 6,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "AI2 ARC",
      "implemented": null
    },
    {
      "name": "Okapi TruthfulQA",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/jon-tow/okapi_truthfulqa/tree/main/data",
      "n_languages": 31,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "TruthfulQA",
      "implemented": null
    },
    {
      "name": "Uhura TruthfulQA",
      "author": "Masakhane",
      "author_url": "https://www.masakhane.io",
      "url": "https://huggingface.co/datasets/masakhane/uhura-truthfulqa",
      "n_languages": 6,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "TruthfulQA",
      "implemented": null
    },
    {
      "name": "XNLI",
      "author": "Meta",
      "author_url": "https://ai.meta.com",
      "url": "https://huggingface.co/datasets/facebook/xnli",
      "n_languages": 14,
      "tasks": [
        "classification"
      ],
      "parallel": 1.0,
      "base": "XNLI",
      "implemented": null
    },
    {
      "name": "AfriXNLI",
      "author": "Masakhane",
      "author_url": "https://www.masakhane.io",
      "url": "https://huggingface.co/datasets/masakhane/afrixnli",
      "n_languages": 18,
      "tasks": [
        "classification"
      ],
      "parallel": 1.0,
      "base": "XNLI",
      "implemented": null
    },
    {
      "name": "Okapi HellaSwag",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/jon-tow/okapi_hellaswag",
      "n_languages": 31,
      "tasks": [
        "question_answering"
      ],
      "parallel": 1.0,
      "base": "HellaSwag",
      "implemented": null
    },
    {
      "name": "WikiANN / PAN-X",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/unimelb-nlp/wikiann",
      "n_languages": 176,
      "tasks": [
        "ner"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "MSVAMP",
      "author": "Microsoft",
      "author_url": "https://microsoft.com",
      "url": "https://huggingface.co/datasets/Mathoctopus/MSVAMP",
      "n_languages": 10,
      "tasks": [
        "math"
      ],
      "parallel": 1.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "XLSUM",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/csebuetnlp/xlsum",
      "n_languages": 45,
      "tasks": [
        "summarization"
      ],
      "parallel": 1.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "SEA-IFEVAL",
      "author": "AI Singapore",
      "author_url": "https://aisingapore.org",
      "url": "https://huggingface.co/datasets/aisingapore/instruction_following-ifeval",
      "n_languages": 7,
      "tasks": [
        "instruction_following"
      ],
      "parallel": 1.0,
      "base": "IFEVAL",
      "implemented": null
    },
    {
      "name": "XTREME",
      "author": "Google",
      "author_url": "https://google.com",
      "url": "https://huggingface.co/datasets/google/xtreme",
      "n_languages": 40,
      "tasks": [
        "translation",
        "classification",
        "question_answering",
        "ner"
      ],
      "parallel": null,
      "base": null,
      "implemented": null
    },
    {
      "name": "XGLUE",
      "author": "Microsoft",
      "author_url": "https://microsoft.com",
      "url": "https://huggingface.co/datasets/microsoft/xglue",
      "n_languages": 18,
      "tasks": [
        "pos"
      ],
      "parallel": null,
      "base": "GLUE",
      "implemented": null
    },
    {
      "name": "IndicGLUE",
      "author": "AI4Bharat",
      "author_url": "https://models.ai4bharat.org",
      "url": "https://huggingface.co/datasets/ai4bharat/indic_glue",
      "n_languages": 11,
      "tasks": [
        "question_answering"
      ],
      "parallel": null,
      "base": "GLUE",
      "implemented": null
    },
    {
      "name": "Opus Gnome",
      "author": "Helsinki NLP",
      "author_url": null,
      "url": "https://huggingface.co/datasets/Helsinki-NLP/opus_gnome",
      "n_languages": 187,
      "tasks": [
        "translation"
      ],
      "parallel": 1.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "Opus Paracrawl",
      "author": "Helsinki NLP",
      "author_url": null,
      "url": "https://huggingface.co/datasets/Helsinki-NLP/opus_paracrawl",
      "n_languages": 43,
      "tasks": [
        "translation"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "CCAligned",
      "author": "Meta",
      "author_url": "https://ai.meta.com",
      "url": "https://huggingface.co/datasets/ahelk/ccaligned_multilingual",
      "n_languages": 137,
      "tasks": [
        "translation"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "OPUS Collection",
      "author": "Helsinki NLP",
      "author_url": null,
      "url": "https://opus.nlpl.eu",
      "n_languages": 747,
      "tasks": [
        "translation"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "MasakhaNER",
      "author": "Masakhane",
      "author_url": "https://www.masakhane.io",
      "url": "https://huggingface.co/datasets/masakhane/masakhaner",
      "n_languages": 10,
      "tasks": [
        "ner"
      ],
      "parallel": null,
      "base": null,
      "implemented": null
    },
    {
      "name": "Multilingual Sentiments",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/tyqiangz/multilingual-sentiments",
      "n_languages": 12,
      "tasks": [
        "sentiment_analysis"
      ],
      "parallel": null,
      "base": null,
      "implemented": null
    },
    {
      "name": "CulturaX",
      "author": "Academic",
      "author_url": null,
      "url": "https://huggingface.co/datasets/uonlp/CulturaX",
      "n_languages": 167,
      "tasks": [
        "language_modeling"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "Tülu 3 SFT Mixture",
      "author": "AllenAI",
      "author_url": "https://allenai.org",
      "url": "https://huggingface.co/datasets/allenai/tulu-3-sft-mixture",
      "n_languages": 70,
      "tasks": [
        "instruction_following"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "xP3",
      "author": "BigScience",
      "author_url": "https://bigscience.huggingface.co",
      "url": "https://huggingface.co/datasets/bigscience/xP3",
      "n_languages": 46,
      "tasks": [
        "instruction_following"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    },
    {
      "name": "Aya",
      "author": "Cohere",
      "author_url": "https://cohere.com",
      "url": "https://huggingface.co/datasets/CohereForAI/aya_dataset",
      "n_languages": 65,
      "tasks": [
        "instruction_following"
      ],
      "parallel": null,
      "base": null,
      "implemented": null
    },
    {
      "name": "Lanfrica",
      "author": "Lanfrica",
      "author_url": "https://lanfrica.com",
      "url": "https://lanfrica.com/records?language=yor&task=machine%20translation",
      "n_languages": 2200,
      "tasks": [
        "datasets"
      ],
      "parallel": null,
      "base": null,
      "implemented": null
    },
    {
      "name": "HuggingFace Languages",
      "author": "HuggingFace",
      "author_url": "https://huggingface.co",
      "url": "https://huggingface.co/languages",
      "n_languages": 4680,
      "tasks": [
        "datasets",
        "models"
      ],
      "parallel": null,
      "base": null,
      "implemented": null
    },
    {
      "name": "HuggingFace Multilingual Datasets",
      "author": "HuggingFace",
      "author_url": "https://huggingface.co",
      "url": "https://huggingface.co/datasets?other=multilinguality:multilingual",
      "n_languages": 2012,
      "tasks": [
        "datasets"
      ],
      "parallel": 0.0,
      "base": null,
      "implemented": null
    }
  ]
}