diff --git "a/frontend/public/results.json" "b/frontend/public/results.json"
--- "a/frontend/public/results.json"
+++ "b/frontend/public/results.json"
@@ -194,18 +194,6 @@
       "translation_bleu": 0.37,
       "translation_chrf": 0.49
     },
-    {
-      "language_name": "French",
-      "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": "Chinese",
       "speakers": 1304678914,
@@ -230,18 +218,6 @@
       "translation_bleu": 0.32,
       "translation_chrf": 0.49
     },
-    {
-      "language_name": "Portuguese",
-      "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": "Spanish",
       "speakers": 493528077,
@@ -254,6 +230,18 @@
       "translation_bleu": 0.28,
       "translation_chrf": 0.46
     },
+    {
+      "language_name": "Arabic",
+      "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",
       "speakers": 290790290,
@@ -267,29 +255,17 @@
       "translation_chrf": 0.42
     },
     {
-      "language_name": "Punjabi",
-      "speakers": 203571210,
+      "language_name": "French",
+      "speakers": 278611507,
       "family": "Indo-European",
-      "average": 0.42,
+      "average": 0.46,
       "in_benchmark": true,
       "NaN": 0.0,
-      "classification_accuracy": 0.4,
-      "language_modeling_chrf": 0.87,
-      "translation_bleu": 0.34,
+      "classification_accuracy": 0.53,
+      "language_modeling_chrf": 0.98,
+      "translation_bleu": 0.32,
       "translation_chrf": 0.49
     },
-    {
-      "language_name": "Arabic",
-      "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": "Bangla",
       "speakers": 267193288,
@@ -303,11 +279,35 @@
       "translation_chrf": 0.41
     },
     {
-      "language_name": "Afar",
-      "speakers": 2119663,
-      "family": "Afro-Asiatic",
+      "language_name": "Portuguese",
+      "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",
+      "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",
+      "speakers": 195841151,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -315,11 +315,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Okanagan",
-      "speakers": 490,
-      "family": "Salishan",
+      "language_name": "Swahili",
+      "speakers": 171610296,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -327,11 +327,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nzima",
-      "speakers": 293402,
-      "family": "Atlantic-Congo",
+      "language_name": "Indonesian",
+      "speakers": 171207687,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -339,8 +339,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Occitan",
-      "speakers": 2040398,
+      "language_name": "German",
+      "speakers": 136350226,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -351,11 +351,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ojibwa",
-      "speakers": 23747,
-      "family": "Algic",
+      "language_name": "Japanese",
+      "speakers": 119729026,
+      "family": "Japonic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -363,11 +363,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Oji-Cree",
-      "speakers": 15078,
-      "family": "Algic",
+      "language_name": "Telugu",
+      "speakers": 95478480,
+      "family": "Dravidian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -375,11 +375,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Odia",
-      "speakers": 42434880,
+      "language_name": "Western Panjabi",
+      "speakers": 93433552,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -387,9 +387,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Oromo",
-      "speakers": 34897121,
-      "family": "Afro-Asiatic",
+      "language_name": "Marathi",
+      "speakers": 92826300,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -399,11 +399,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nyamwezi",
-      "speakers": 1932242,
-      "family": "Atlantic-Congo",
+      "language_name": "Javanese",
+      "speakers": 91180665,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -411,11 +411,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ossetic",
-      "speakers": 541444,
-      "family": "Indo-European",
+      "language_name": "Vietnamese",
+      "speakers": 86222962,
+      "family": "Austroasiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -423,11 +423,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Osage",
-      "speakers": 0,
-      "family": "Siouan",
+      "language_name": "Tamil",
+      "speakers": 85616159,
+      "family": "Dravidian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -435,9 +435,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pangasinan",
-      "speakers": 1528534,
-      "family": "Austronesian",
+      "language_name": "Persian",
+      "speakers": 84710459,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -447,9 +447,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pampanga",
-      "speakers": 2511163,
-      "family": "Austronesian",
+      "language_name": "Wu Chinese",
+      "speakers": 83641200,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -459,9 +459,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Papiamento",
-      "speakers": 211640,
-      "family": "Indo-European",
+      "language_name": "Turkish",
+      "speakers": 80360704,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -471,11 +471,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Palauan",
-      "speakers": 16047,
-      "family": "Austronesian",
+      "language_name": "Cantonese",
+      "speakers": 79654759,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -483,11 +483,23 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Picard",
-      "speakers": 746330,
+      "language_name": "Korean",
+      "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",
+      "speakers": 70247060,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -495,11 +507,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nyankole",
-      "speakers": 2724939,
-      "family": "Atlantic-Congo",
+      "language_name": "Filipino",
+      "speakers": 67471096,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -507,11 +519,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Navajo",
-      "speakers": 166320,
-      "family": "Athabaskan-Eyak-Tlingit",
+      "language_name": "Egyptian Arabic",
+      "speakers": 66639360,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -519,9 +531,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nyanja",
-      "speakers": 17026781,
-      "family": "Atlantic-Congo",
+      "language_name": "Gujarati",
+      "speakers": 61721799,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -531,11 +543,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ngiemboon",
-      "speakers": 388430,
-      "family": "Atlantic-Congo",
+      "language_name": "Thai",
+      "speakers": 55181920,
+      "family": "Tai-Kadai",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -543,9 +555,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Huasteca Nahuatl",
-      "speakers": 501735,
-      "family": "Uto-Aztecan",
+      "language_name": "Pashto",
+      "speakers": 53542641,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -555,11 +567,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ngaju",
-      "speakers": 987996,
-      "family": "Austronesian",
+      "language_name": "Kannada",
+      "speakers": 49065330,
+      "family": "Dravidian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -567,9 +579,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Niuean",
-      "speakers": 1120,
-      "family": "Austronesian",
+      "language_name": "Nigerian Pidgin",
+      "speakers": 44945880,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -579,11 +591,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ao Naga",
-      "speakers": 305001,
-      "family": "Sino-Tibetan",
+      "language_name": "Malayalam",
+      "speakers": 43257484,
+      "family": "Dravidian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -591,8 +603,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dutch",
-      "speakers": 31765645,
+      "language_name": "Odia",
+      "speakers": 42434880,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -603,9 +615,21 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kwasio",
-      "speakers": 8878,
-      "family": "Atlantic-Congo",
+      "language_name": "Polish",
+      "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",
+      "speakers": 40426580,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -615,9 +639,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Norwegian Nynorsk",
-      "speakers": 1366860,
-      "family": "Indo-European",
+      "language_name": "Hausa",
+      "speakers": 40411882,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -627,11 +651,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Norwegian",
-      "speakers": 5467440,
+      "language_name": "Sindhi",
+      "speakers": 40329510,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -639,9 +663,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Naxi",
-      "speakers": 334565,
-      "family": "Sino-Tibetan",
+      "language_name": "North Levantine Arabic",
+      "speakers": 39031474,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -651,11 +675,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern Thai",
-      "speakers": 6621830,
-      "family": "Tai-Kadai",
+      "language_name": "Malay",
+      "speakers": 38097307,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -663,11 +687,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nimadi",
-      "speakers": 1723917,
-      "family": "Indo-European",
+      "language_name": "Burmese",
+      "speakers": 36559231,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -675,9 +699,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "N’Ko",
-      "speakers": 626370,
-      "family": "Artificial Language",
+      "language_name": "Amharic",
+      "speakers": 35728475,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -687,9 +711,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "South Ndebele",
-      "speakers": 903418,
-      "family": "Atlantic-Congo",
+      "language_name": "Algerian Arabic",
+      "speakers": 35667507,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -699,11 +723,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Naskapi",
-      "speakers": 1395,
-      "family": "Algic",
+      "language_name": "Oromo",
+      "speakers": 34897121,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -711,9 +735,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern Sotho",
-      "speakers": 5307578,
-      "family": "Atlantic-Congo",
+      "language_name": "Bhojpuri",
+      "speakers": 32934797,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -723,9 +747,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nuer",
-      "speakers": 591427,
-      "family": "Nilotic",
+      "language_name": "Uzbek",
+      "speakers": 32792780,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -735,11 +759,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nigerian Pidgin",
-      "speakers": 44945880,
-      "family": "Indo-European",
+      "language_name": "Azerbaijani",
+      "speakers": 32446682,
+      "family": "Turkic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -747,9 +771,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pijin",
-      "speakers": 561780,
-      "family": "Indo-European",
+      "language_name": "Hakka Chinese",
+      "speakers": 32062460,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -759,11 +783,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pennsylvania German",
-      "speakers": 129729,
-      "family": "Indo-European",
+      "language_name": "Sundanese",
+      "speakers": 32043120,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -771,11 +795,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rohingya",
-      "speakers": 1824082,
+      "language_name": "Dutch",
+      "speakers": 31765645,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -783,11 +807,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Riffian",
-      "speakers": 3692411,
+      "language_name": "Moroccan Arabic",
+      "speakers": 30938679,
       "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -795,11 +819,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rajbanshi",
-      "speakers": 133443,
+      "language_name": "Ukrainian",
+      "speakers": 29348975,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -807,11 +831,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rangpuri",
-      "speakers": 16274502,
-      "family": "Indo-European",
+      "language_name": "Yoruba",
+      "speakers": 28685568,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -819,8 +843,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Romansh",
-      "speakers": 42020,
+      "language_name": "Saraiki",
+      "speakers": 28020120,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -831,11 +855,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kalo Finnish Romani",
-      "speakers": 5015,
-      "family": "Indo-European",
+      "language_name": "Igbo",
+      "speakers": 27823640,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -843,9 +867,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sinte Romani",
-      "speakers": 24372,
-      "family": "Indo-European",
+      "language_name": "Min Nan Chinese",
+      "speakers": 26486380,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -855,11 +879,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Domari",
-      "speakers": 1613543,
-      "family": "Indo-European",
+      "language_name": "Cebuano",
+      "speakers": 26203440,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -867,11 +891,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tavringer Romani",
-      "speakers": 9488,
-      "family": "Speech Register",
+      "language_name": "Awadhi",
+      "speakers": 25862924,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -879,9 +903,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rundi",
-      "speakers": 7475454,
-      "family": "Atlantic-Congo",
+      "language_name": "Malagasy",
+      "speakers": 24260130,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -891,9 +915,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ronga",
-      "speakers": 1023339,
-      "family": "Atlantic-Congo",
+      "language_name": "Gan Chinese",
+      "speakers": 23698340,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -915,9 +939,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tae'",
-      "speakers": 293729,
-      "family": "Austronesian",
+      "language_name": "Bavarian",
+      "speakers": 22043627,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -927,11 +951,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rombo",
-      "speakers": 433291,
-      "family": "Atlantic-Congo",
+      "language_name": "Nepali",
+      "speakers": 20903374,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -939,11 +963,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rotuman",
-      "speakers": 2527,
-      "family": "Austronesian",
+      "language_name": "Maithili",
+      "speakers": 19249149,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -951,8 +975,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Russian",
-      "speakers": 195841151,
+      "language_name": "Assamese",
+      "speakers": 17239170,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -963,11 +987,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rusyn",
-      "speakers": 527075,
-      "family": "Indo-European",
+      "language_name": "Nyanja",
+      "speakers": 17026781,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -975,11 +999,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Roviana",
-      "speakers": 9591,
-      "family": "Austronesian",
+      "language_name": "Somali",
+      "speakers": 16911645,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -987,9 +1011,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Riang (India)",
-      "speakers": 172392,
-      "family": "Sino-Tibetan",
+      "language_name": "Madurese",
+      "speakers": 16822638,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -999,9 +1023,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Romagnol",
-      "speakers": 0,
-      "family": "Indo-European",
+      "language_name": "Northeastern Thai",
+      "speakers": 16554576,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1011,8 +1035,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Plautdietsch",
-      "speakers": 90466,
+      "language_name": "Rangpuri",
+      "speakers": 16274502,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1023,11 +1047,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rejang",
-      "speakers": 1228320,
-      "family": "Austronesian",
+      "language_name": "Magahi",
+      "speakers": 15913080,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1035,8 +1059,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Palatine German",
-      "speakers": 0,
+      "language_name": "Haryanvi",
+      "speakers": 15913080,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1047,9 +1071,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lomwe",
-      "speakers": 2046678,
-      "family": "Atlantic-Congo",
+      "language_name": "Marwari",
+      "speakers": 15913080,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1059,11 +1083,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pökoot",
-      "speakers": 369343,
-      "family": "Nilotic",
+      "language_name": "Serbian",
+      "speakers": 15602410,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1071,8 +1095,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Polish",
-      "speakers": 41077399,
+      "language_name": "Sinhala",
+      "speakers": 15564656,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -1083,11 +1107,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Piedmontese",
-      "speakers": 6178,
-      "family": "Indo-European",
+      "language_name": "Khmer",
+      "speakers": 15065030,
+      "family": "Austroasiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1095,11 +1119,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pontic",
-      "speakers": 392463,
+      "language_name": "Chhattisgarhi",
+      "speakers": 14586990,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1107,11 +1131,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pohnpeian",
-      "speakers": 23560,
-      "family": "Austronesian",
+      "language_name": "Nigerian Fulfulde",
+      "speakers": 14339876,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1119,11 +1143,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Maliseet-Passamaquoddy",
-      "speakers": 490,
-      "family": "Algic",
+      "language_name": "Zulu",
+      "speakers": 13973830,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1131,11 +1155,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Parsi-Dari",
-      "speakers": 864342,
-      "family": "Bookkeeping",
+      "language_name": "Kazakh",
+      "speakers": 13637392,
+      "family": "Turkic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1143,8 +1167,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Prussian",
-      "speakers": 38,
+      "language_name": "Deccan",
+      "speakers": 13128291,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1155,11 +1179,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pashto",
-      "speakers": 53542641,
+      "language_name": "Czech",
+      "speakers": 13045532,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1167,11 +1191,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Punu",
-      "speakers": 200782,
-      "family": "Atlantic-Congo",
+      "language_name": "Swedish",
+      "speakers": 12932871,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1179,11 +1203,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Quechua",
-      "speakers": 11385851,
-      "family": "Quechuan",
+      "language_name": "Hungarian",
+      "speakers": 12443430,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1191,11 +1215,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kʼicheʼ",
-      "speakers": 1200731,
-      "family": "Mayan",
+      "language_name": "Greek",
+      "speakers": 12292242,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1203,11 +1227,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chimborazo Highland Quichua",
-      "speakers": 963579,
-      "family": "Quechuan",
+      "language_name": "Shona",
+      "speakers": 11782503,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1215,8 +1239,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rajasthani",
-      "speakers": 1326090,
+      "language_name": "Low German",
+      "speakers": 11520008,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1227,9 +1251,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Réunion Creole French",
-      "speakers": 559185,
-      "family": "Indo-European",
+      "language_name": "Akan",
+      "speakers": 11442678,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1239,9 +1263,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Eastern Huasteca Nahuatl",
-      "speakers": 501735,
-      "family": "Uto-Aztecan",
+      "language_name": "Quechua",
+      "speakers": 11385851,
+      "family": "Quechuan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1251,11 +1275,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Newari",
-      "speakers": 1000821,
-      "family": "Sino-Tibetan",
+      "language_name": "Central Kurdish",
+      "speakers": 11086549,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1263,11 +1287,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ndonga",
-      "speakers": 552315,
+      "language_name": "Kinyarwanda",
+      "speakers": 11083625,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1275,11 +1299,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mende",
-      "speakers": 1813083,
-      "family": "Mande",
+      "language_name": "Wolof",
+      "speakers": 11025494,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1287,9 +1311,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Maithili",
-      "speakers": 19249149,
-      "family": "Indo-European",
+      "language_name": "Tunisian Arabic",
+      "speakers": 10549080,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1299,11 +1323,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Makasar",
-      "speakers": 1949290,
+      "language_name": "Iloko",
+      "speakers": 10481376,
       "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1311,11 +1335,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mandingo",
-      "speakers": 3511762,
-      "family": "Mande",
+      "language_name": "Xhosa",
+      "speakers": 10182944,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1323,11 +1347,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Masai",
-      "speakers": 1734738,
-      "family": "Nilotic",
+      "language_name": "Tigrinya",
+      "speakers": 10145911,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1335,11 +1359,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central Mazahua",
-      "speakers": 437410,
-      "family": "Otomanguean",
+      "language_name": "Belarusian",
+      "speakers": 10064517,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1347,11 +1371,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Moksha",
-      "speakers": 297616,
-      "family": "Uralic",
+      "language_name": "Luba-Lulua",
+      "speakers": 9770880,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1359,11 +1383,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Maguindanaon",
-      "speakers": 1310172,
-      "family": "Austronesian",
+      "language_name": "Tajik",
+      "speakers": 9644223,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1371,11 +1395,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mandar",
-      "speakers": 245664,
-      "family": "Austronesian",
+      "language_name": "Umbundu",
+      "speakers": 9431467,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1383,11 +1407,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Meru",
-      "speakers": 2141116,
-      "family": "Atlantic-Congo",
+      "language_name": "Bambara",
+      "speakers": 9385632,
+      "family": "Mande",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1395,11 +1419,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rwa",
-      "speakers": 128816,
-      "family": "Atlantic-Congo",
+      "language_name": "Afrikaans",
+      "speakers": 9318845,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1407,9 +1431,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hassaniyya",
-      "speakers": 7239,
-      "family": "Afro-Asiatic",
+      "language_name": "Hiligaynon",
+      "speakers": 9171204,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1419,11 +1443,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Pattani Malay",
-      "speakers": 3448870,
-      "family": "Austronesian",
+      "language_name": "Kikuyu",
+      "speakers": 9099743,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1431,11 +1455,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Morisyen",
-      "speakers": 1241433,
+      "language_name": "Haitian Creole",
+      "speakers": 8964918,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1443,11 +1467,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mandjak",
-      "speakers": 121170,
-      "family": "Atlantic-Congo",
+      "language_name": "Catalan",
+      "speakers": 8679139,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1455,9 +1479,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Malagasy",
-      "speakers": 24260130,
-      "family": "Austronesian",
+      "language_name": "Hebrew",
+      "speakers": 8675480,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1467,9 +1491,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Makhuwa-Meetto",
-      "speakers": 1354419,
-      "family": "Atlantic-Congo",
+      "language_name": "Sichuan Yi",
+      "speakers": 8364120,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1479,11 +1503,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Metaʼ",
-      "speakers": 130401,
+      "language_name": "Mossi",
+      "speakers": 8334160,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1491,9 +1515,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Eastern Magar",
-      "speakers": 333607,
-      "family": "Sino-Tibetan",
+      "language_name": "Baluchi",
+      "speakers": 8227887,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1503,11 +1527,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Magahi",
-      "speakers": 15913080,
+      "language_name": "Sylheti",
+      "speakers": 8132550,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1515,11 +1539,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mafa",
-      "speakers": 205313,
-      "family": "Afro-Asiatic",
+      "language_name": "Kimbundu",
+      "speakers": 8130575,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1527,11 +1551,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Madurese",
-      "speakers": 16822638,
-      "family": "Austronesian",
+      "language_name": "Uyghur",
+      "speakers": 8052967,
+      "family": "Turkic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1539,11 +1563,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Laz",
-      "speakers": 22965,
-      "family": "Kartvelian",
+      "language_name": "Minangkabau",
+      "speakers": 8010780,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1551,8 +1575,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lambadi",
-      "speakers": 3580443,
+      "language_name": "Swiss German",
+      "speakers": 7956952,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1563,8 +1587,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lombard",
-      "speakers": 3901518,
+      "language_name": "Bulgarian",
+      "speakers": 7878315,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -1575,9 +1599,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lingala",
-      "speakers": 3514491,
-      "family": "Atlantic-Congo",
+      "language_name": "Standard Moroccan Tamazight",
+      "speakers": 7823574,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1587,11 +1611,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lao",
-      "speakers": 5138706,
-      "family": "Tai-Kadai",
+      "language_name": "Fula",
+      "speakers": 7788904,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1599,11 +1623,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mongo",
-      "speakers": 620858,
-      "family": "Atlantic-Congo",
+      "language_name": "Bosnian",
+      "speakers": 7594468,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1611,11 +1635,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lozi",
-      "speakers": 1045596,
+      "language_name": "Rundi",
+      "speakers": 7475454,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1623,8 +1647,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern Luri",
-      "speakers": 2020512,
+      "language_name": "Kanauji",
+      "speakers": 7426104,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1635,9 +1659,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lithuanian",
-      "speakers": 2488617,
-      "family": "Indo-European",
+      "language_name": "Santali",
+      "speakers": 7293495,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1647,8 +1671,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Latgalian",
-      "speakers": 167429,
+      "language_name": "Danish",
+      "speakers": 7072056,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -1659,11 +1683,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Luba-Katanga",
-      "speakers": 2340940,
-      "family": "Atlantic-Congo",
+      "language_name": "Turkmen",
+      "speakers": 6870838,
+      "family": "Turkic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1671,9 +1695,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Luba-Lulua",
-      "speakers": 9770880,
-      "family": "Atlantic-Congo",
+      "language_name": "Kurdish",
+      "speakers": 6866757,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1683,9 +1707,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Luo (Kenya and Tanzania)",
-      "speakers": 5245734,
-      "family": "Nilotic",
+      "language_name": "Croatian",
+      "speakers": 6813164,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1695,11 +1719,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Luyia",
-      "speakers": 5888069,
-      "family": "Atlantic-Congo",
+      "language_name": "Albanian",
+      "speakers": 6791906,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1707,11 +1731,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Southern Luri",
-      "speakers": 1019080,
+      "language_name": "Slovak",
+      "speakers": 6680269,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1719,9 +1743,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Latvian",
-      "speakers": 1147550,
-      "family": "Indo-European",
+      "language_name": "Dyula",
+      "speakers": 6667328,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1731,9 +1755,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Eastern Lawa",
-      "speakers": 6898,
-      "family": "Austroasiatic",
+      "language_name": "Northern Thai",
+      "speakers": 6621830,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1743,11 +1767,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Literary Chinese",
-      "speakers": 0,
-      "family": "Sino-Tibetan",
+      "language_name": "Mongolian",
+      "speakers": 6572846,
+      "family": "Mongolic-Khitan",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1755,11 +1779,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mbunga",
-      "speakers": 819739,
+      "language_name": "Southern Sotho",
+      "speakers": 6390567,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1767,9 +1791,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Marshallese",
-      "speakers": 56879,
-      "family": "Austronesian",
+      "language_name": "Krio",
+      "speakers": 6293684,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1779,11 +1803,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Māori",
-      "speakers": 137913,
-      "family": "Austronesian",
+      "language_name": "Tachelhit",
+      "speakers": 6187736,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1791,11 +1815,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Indus Kohistani",
-      "speakers": 326901,
-      "family": "Indo-European",
+      "language_name": "Tswana",
+      "speakers": 6113428,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1803,8 +1827,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Marwari",
-      "speakers": 15913080,
+      "language_name": "Mewati",
+      "speakers": 6100014,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1815,9 +1839,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mentawai",
-      "speakers": 64086,
-      "family": "Austronesian",
+      "language_name": "Luyia",
+      "speakers": 5888069,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1827,11 +1851,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Manyika",
-      "speakers": 945510,
-      "family": "Atlantic-Congo",
+      "language_name": "Guarani",
+      "speakers": 5827107,
+      "family": "Tupian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1839,9 +1863,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Burmese",
-      "speakers": 36559231,
-      "family": "Sino-Tibetan",
+      "language_name": "Finnish",
+      "speakers": 5736842,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -1851,11 +1875,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Erzya",
-      "speakers": 439338,
-      "family": "Uralic",
+      "language_name": "Ganda",
+      "speakers": 5622890,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1863,9 +1887,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Masaaba",
-      "speakers": 1254337,
-      "family": "Atlantic-Congo",
+      "language_name": "Betawi",
+      "speakers": 5607546,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1875,11 +1899,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mazanderani",
-      "speakers": 4246165,
+      "language_name": "Kashmiri",
+      "speakers": 5598085,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1887,9 +1911,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nauru",
-      "speakers": 6930,
-      "family": "Austronesian",
+      "language_name": "Southern Thai",
+      "speakers": 5518192,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -1899,11 +1923,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Min Nan Chinese",
-      "speakers": 26486380,
-      "family": "Sino-Tibetan",
+      "language_name": "Norwegian Bokmål",
+      "speakers": 5468932,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1911,8 +1935,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Neapolitan",
-      "speakers": 605306,
+      "language_name": "Norwegian",
+      "speakers": 5467440,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1923,11 +1947,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nama",
-      "speakers": 289308,
-      "family": "Khoe-Kwadi",
+      "language_name": "Bemba",
+      "speakers": 5402246,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1935,8 +1959,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Norwegian Bokmål",
-      "speakers": 5468932,
+      "language_name": "Armenian",
+      "speakers": 5317273,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -1947,11 +1971,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central Huasteca Nahuatl",
-      "speakers": 244435,
-      "family": "Uto-Aztecan",
+      "language_name": "Northern Sotho",
+      "speakers": 5307578,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1959,11 +1983,35 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "North Ndebele",
-      "speakers": 1745556,
-      "family": "Atlantic-Congo",
+      "language_name": "Luo (Kenya and Tanzania)",
+      "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",
+      "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",
+      "speakers": 5138706,
+      "family": "Tai-Kadai",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -1971,8 +2019,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ndau",
-      "speakers": 3867046,
+      "language_name": "Sukuma",
+      "speakers": 5094094,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -1983,8 +2031,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Low German",
-      "speakers": 11520008,
+      "language_name": "Konkani",
+      "speakers": 4906533,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -1995,9 +2043,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nepali",
-      "speakers": 20903374,
-      "family": "Indo-European",
+      "language_name": "Tsonga",
+      "speakers": 4880932,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2007,9 +2055,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kita Maninkakan",
-      "speakers": 977670,
-      "family": "Mande",
+      "language_name": "Main-Franconian",
+      "speakers": 4809582,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2019,11 +2067,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Muscogee",
-      "speakers": 3992,
-      "family": "Muskogean",
+      "language_name": "Balinese",
+      "speakers": 4806468,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2031,11 +2079,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mi'kmaw",
-      "speakers": 7916,
-      "family": "Algic",
+      "language_name": "Ewe",
+      "speakers": 4690857,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2043,9 +2091,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mundang",
-      "speakers": 277450,
-      "family": "Atlantic-Congo",
+      "language_name": "Zhuang",
+      "speakers": 4321462,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2055,8 +2103,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Minangkabau",
-      "speakers": 8010780,
+      "language_name": "Buginese",
+      "speakers": 4298211,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": true,
@@ -2067,11 +2115,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Macedonian",
-      "speakers": 1608565,
+      "language_name": "Mazanderani",
+      "speakers": 4246165,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2079,9 +2127,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Malayalam",
-      "speakers": 43257484,
-      "family": "Dravidian",
+      "language_name": "Goan Konkani",
+      "speakers": 4243488,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2091,11 +2139,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Masalit",
-      "speakers": 451060,
-      "family": "Maban",
+      "language_name": "Kamba",
+      "speakers": 4068120,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2103,11 +2151,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mongolian",
-      "speakers": 6572846,
-      "family": "Mongolic-Khitan",
+      "language_name": "Kalenjin",
+      "speakers": 4068120,
+      "family": "Nilotic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2115,9 +2163,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Manipuri",
-      "speakers": 1476591,
-      "family": "Sino-Tibetan",
+      "language_name": "Banjar",
+      "speakers": 4010288,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2127,9 +2175,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mon",
-      "speakers": 966114,
-      "family": "Austroasiatic",
+      "language_name": "Northern Hindko",
+      "speakers": 3969517,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2139,9 +2187,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Innu-aimun",
-      "speakers": 12062,
-      "family": "Algic",
+      "language_name": "Makhuwa",
+      "speakers": 3912766,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2151,9 +2199,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mohawk",
-      "speakers": 1772,
-      "family": "Iroquoian",
+      "language_name": "Gilaki",
+      "speakers": 3906472,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2163,20 +2211,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "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": "Marathi",
-      "speakers": 92826300,
+      "language_name": "Lombard",
+      "speakers": 3901518,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -2187,9 +2223,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Magar",
-      "speakers": 251722,
-      "family": "Sino-Tibetan",
+      "language_name": "Zarma",
+      "speakers": 3871308,
+      "family": "Songhay",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2199,9 +2235,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Mari",
-      "speakers": 29762,
-      "family": "Uralic",
+      "language_name": "Ndau",
+      "speakers": 3867046,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2211,9 +2247,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mru",
-      "speakers": 29277,
-      "family": "Sino-Tibetan",
+      "language_name": "Sidamo",
+      "speakers": 3783955,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2223,8 +2259,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Malay",
-      "speakers": 38097307,
+      "language_name": "Achinese",
+      "speakers": 3738364,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": true,
@@ -2235,11 +2271,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Maltese",
-      "speakers": 457267,
-      "family": "Afro-Asiatic",
+      "language_name": "Shekhawati",
+      "speakers": 3713052,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2247,9 +2283,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mewari",
-      "speakers": 1286307,
-      "family": "Indo-European",
+      "language_name": "Riffian",
+      "speakers": 3692411,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2259,9 +2295,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kinyarwanda",
-      "speakers": 11083625,
-      "family": "Atlantic-Congo",
+      "language_name": "Shan",
+      "speakers": 3687984,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2271,9 +2307,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Safaliba",
-      "speakers": 4108,
-      "family": "Atlantic-Congo",
+      "language_name": "Lambadi",
+      "speakers": 3580443,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2283,9 +2319,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central Okinawan",
-      "speakers": 966404,
-      "family": "Japonic",
+      "language_name": "Garhwali",
+      "speakers": 3580443,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2295,11 +2331,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Muslim Tat",
-      "speakers": 22453,
-      "family": "Indo-European",
+      "language_name": "Georgian",
+      "speakers": 3543646,
+      "family": "Kartvelian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2307,11 +2343,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tuvalu",
-      "speakers": 9868,
-      "family": "Austronesian",
+      "language_name": "Galician",
+      "speakers": 3515530,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2319,11 +2355,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tasawaq",
-      "speakers": 7970,
-      "family": "Songhay",
+      "language_name": "Lingala",
+      "speakers": 3514491,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2331,9 +2367,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tahitian",
-      "speakers": 91488,
-      "family": "Austronesian",
+      "language_name": "Mandingo",
+      "speakers": 3511762,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2343,9 +2379,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tuvinian",
-      "speakers": 184239,
-      "family": "Turkic",
+      "language_name": "Central Atlas Tamazight",
+      "speakers": 3485047,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2355,9 +2391,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central Atlas Tamazight",
-      "speakers": 3485047,
-      "family": "Afro-Asiatic",
+      "language_name": "Pattani Malay",
+      "speakers": 3448870,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2367,9 +2403,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Udmurt",
-      "speakers": 538544,
-      "family": "Uralic",
+      "language_name": "Tiv",
+      "speakers": 3424448,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2379,9 +2415,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Uyghur",
-      "speakers": 8052967,
-      "family": "Turkic",
+      "language_name": "Kabyle",
+      "speakers": 3351886,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2391,9 +2427,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ukrainian",
-      "speakers": 29348975,
-      "family": "Indo-European",
+      "language_name": "Kyrgyz",
+      "speakers": 3338267,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2403,8 +2439,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ulithian",
-      "speakers": 2971,
+      "language_name": "Bikol",
+      "speakers": 3275430,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
@@ -2415,8 +2451,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Umbundu",
-      "speakers": 9431467,
+      "language_name": "Fon",
+      "speakers": 3216150,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": true,
@@ -2427,9 +2463,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mundari",
-      "speakers": 1252287,
-      "family": "Austroasiatic",
+      "language_name": "Gondi",
+      "speakers": 3182616,
+      "family": "Dravidian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2439,11 +2475,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Munda",
-      "speakers": 636523,
-      "family": "Bookkeeping",
+      "language_name": "Waray",
+      "speakers": 3166927,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2451,11 +2487,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Uzbek",
-      "speakers": 32792780,
-      "family": "Turkic",
+      "language_name": "Southern Kurdish",
+      "speakers": 3142162,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2463,9 +2499,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Vai",
-      "speakers": 131906,
-      "family": "Mande",
+      "language_name": "Brahui",
+      "speakers": 3035513,
+      "family": "Dravidian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2475,8 +2511,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Venda",
-      "speakers": 1391759,
+      "language_name": "Baoulé",
+      "speakers": 3022921,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -2487,9 +2523,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Venetian",
-      "speakers": 1380829,
-      "family": "Indo-European",
+      "language_name": "Tibetan",
+      "speakers": 3006697,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2499,9 +2535,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Veps",
-      "speakers": 3543,
-      "family": "Uralic",
+      "language_name": "Ibibio",
+      "speakers": 2996392,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2511,11 +2547,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tumbuka",
-      "speakers": 1780514,
+      "language_name": "Efik",
+      "speakers": 2996392,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2523,11 +2559,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northeastern Thai",
-      "speakers": 16554576,
-      "family": "Tai-Kadai",
+      "language_name": "Sango",
+      "speakers": 2935521,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2535,8 +2571,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Virgin Islands Creole English",
-      "speakers": 3113,
+      "language_name": "Kumaoni",
+      "speakers": 2917398,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -2547,11 +2583,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tooro",
-      "speakers": 821807,
-      "family": "Atlantic-Congo",
+      "language_name": "Aymara",
+      "speakers": 2838620,
+      "family": "Aymaran",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2559,9 +2595,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kathoriya Tharu",
-      "speakers": 72787,
-      "family": "Indo-European",
+      "language_name": "Nyankole",
+      "speakers": 2724939,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2571,8 +2607,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Talysh",
-      "speakers": 1000168,
+      "language_name": "Jamaican Creole English",
+      "speakers": 2668142,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -2583,11 +2619,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tamashek",
-      "speakers": 1776965,
-      "family": "Afro-Asiatic",
+      "language_name": "Dogri",
+      "speakers": 2652180,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2595,11 +2631,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tswana",
-      "speakers": 6113428,
+      "language_name": "Gusii",
+      "speakers": 2622867,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2607,9 +2643,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ménik",
-      "speakers": 3305,
-      "family": "Atlantic-Congo",
+      "language_name": "Sasak",
+      "speakers": 2590152,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2619,9 +2655,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tongan",
-      "speakers": 100790,
-      "family": "Austronesian",
+      "language_name": "Kurukh",
+      "speakers": 2519571,
+      "family": "Dravidian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2631,9 +2667,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nyasa Tonga",
-      "speakers": 207727,
-      "family": "Atlantic-Congo",
+      "language_name": "Pampanga",
+      "speakers": 2511163,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2643,11 +2679,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tok Pisin",
-      "speakers": 5154217,
-      "family": "Indo-European",
+      "language_name": "West Albay Bikol",
+      "speakers": 2511163,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2655,9 +2691,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Turkish",
-      "speakers": 80360704,
-      "family": "Turkic",
+      "language_name": "Lithuanian",
+      "speakers": 2488617,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2667,8 +2703,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Turoyo",
-      "speakers": 3035,
+      "language_name": "Beja",
+      "speakers": 2460326,
       "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
@@ -2679,8 +2715,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Taroko",
-      "speakers": 4721,
+      "language_name": "Batak Toba",
+      "speakers": 2456639,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
@@ -2691,8 +2727,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Torwali",
-      "speakers": 123756,
+      "language_name": "Sadri",
+      "speakers": 2386962,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -2703,21 +2739,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "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": "Tsakonian",
-      "speakers": 202,
-      "family": "Indo-European",
+      "language_name": "Waddar",
+      "speakers": 2386962,
+      "family": "Dravidian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2727,9 +2751,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tausug",
-      "speakers": 1200991,
-      "family": "Austronesian",
+      "language_name": "Luba-Katanga",
+      "speakers": 2340940,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2739,9 +2763,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tshangla",
-      "speakers": 117348,
-      "family": "Sino-Tibetan",
+      "language_name": "Chiga",
+      "speakers": 2335662,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2751,11 +2775,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tatar",
-      "speakers": 1984108,
-      "family": "Turkic",
+      "language_name": "Soga",
+      "speakers": 2292409,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2763,9 +2787,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Vietnamese",
-      "speakers": 86222962,
-      "family": "Austroasiatic",
+      "language_name": "Swati",
+      "speakers": 2212379,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2775,8 +2799,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "West Flemish",
-      "speakers": 1172070,
+      "language_name": "Hazaragi",
+      "speakers": 2161984,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -2787,9 +2811,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tokelau",
-      "speakers": 1285,
-      "family": "Austronesian",
+      "language_name": "Meru",
+      "speakers": 2141116,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2811,9 +2835,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sherpa",
-      "speakers": 157705,
-      "family": "Sino-Tibetan",
+      "language_name": "Afar",
+      "speakers": 2119663,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2823,9 +2847,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yao",
-      "speakers": 722357,
-      "family": "Atlantic-Congo",
+      "language_name": "Teso",
+      "speakers": 2082973,
+      "family": "Nilotic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2835,9 +2859,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yapese",
-      "speakers": 6556,
-      "family": "Austronesian",
+      "language_name": "Lomwe",
+      "speakers": 2046678,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2847,11 +2871,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yangben",
-      "speakers": 2303,
-      "family": "Atlantic-Congo",
+      "language_name": "Occitan",
+      "speakers": 2040398,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2859,9 +2883,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yemba",
-      "speakers": 443920,
-      "family": "Atlantic-Congo",
+      "language_name": "Western Balochi",
+      "speakers": 2037382,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2871,11 +2895,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yiddish",
-      "speakers": 997214,
+      "language_name": "Northern Luri",
+      "speakers": 2020512,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2883,11 +2907,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yoruba",
-      "speakers": 28685568,
-      "family": "Atlantic-Congo",
+      "language_name": "Wagdi",
+      "speakers": 1989135,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2895,9 +2919,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nheengatu",
-      "speakers": 26171,
-      "family": "Tupian",
+      "language_name": "Tulu",
+      "speakers": 1989135,
+      "family": "Dravidian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2907,9 +2931,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yucateco",
-      "speakers": 861955,
-      "family": "Mayan",
+      "language_name": "Khandesi",
+      "speakers": 1989135,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2919,9 +2943,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Cantonese",
-      "speakers": 79654759,
-      "family": "Sino-Tibetan",
+      "language_name": "Tatar",
+      "speakers": 1984108,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2931,11 +2955,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Zhuang",
-      "speakers": 4321462,
-      "family": "Tai-Kadai",
+      "language_name": "Slovenian",
+      "speakers": 1973181,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -2943,9 +2967,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Zaghawa",
-      "speakers": 232364,
-      "family": "Saharan",
+      "language_name": "Makasar",
+      "speakers": 1949290,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2955,9 +2979,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ngazidja Comorian",
-      "speakers": 313124,
-      "family": "Atlantic-Congo",
+      "language_name": "Wolaytta",
+      "speakers": 1946034,
+      "family": "Ta-Ne-Omotic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2967,9 +2991,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Zeelandic",
-      "speakers": 241926,
-      "family": "Indo-European",
+      "language_name": "Nyamwezi",
+      "speakers": 1932242,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -2979,9 +3003,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Standard Moroccan Tamazight",
-      "speakers": 7823574,
-      "family": "Afro-Asiatic",
+      "language_name": "Bodo",
+      "speakers": 1856526,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -2991,8 +3015,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Negeri Sembilan Malay",
-      "speakers": 391825,
+      "language_name": "Lampung Api",
+      "speakers": 1842479,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
@@ -3003,9 +3027,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Zulu",
-      "speakers": 13973830,
-      "family": "Atlantic-Congo",
+      "language_name": "Chuvash",
+      "speakers": 1842386,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -3015,11 +3039,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Soga",
-      "speakers": 2292409,
-      "family": "Atlantic-Congo",
+      "language_name": "Bashkir",
+      "speakers": 1842386,
+      "family": "Turkic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3027,9 +3051,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mingrelian",
-      "speakers": 439670,
-      "family": "Kartvelian",
+      "language_name": "Rohingya",
+      "speakers": 1824082,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3039,9 +3063,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Main-Franconian",
-      "speakers": 4809582,
-      "family": "Indo-European",
+      "language_name": "Mende",
+      "speakers": 1813083,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3051,8 +3075,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Xhosa",
-      "speakers": 10182944,
+      "language_name": "Tumbuka",
+      "speakers": 1780514,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": true,
@@ -3063,9 +3087,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Makhuwa",
-      "speakers": 3912766,
-      "family": "Atlantic-Congo",
+      "language_name": "Tamashek",
+      "speakers": 1776965,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3075,9 +3099,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Votic",
-      "speakers": 0,
-      "family": "Uralic",
+      "language_name": "North Ndebele",
+      "speakers": 1745556,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3087,9 +3111,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Võro",
-      "speakers": 70031,
-      "family": "Uralic",
+      "language_name": "Masai",
+      "speakers": 1734738,
+      "family": "Nilotic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3099,8 +3123,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Vunjo",
-      "speakers": 433291,
+      "language_name": "Serer",
+      "speakers": 1731004,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -3111,8 +3135,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Walloon",
-      "speakers": 679801,
+      "language_name": "Nimadi",
+      "speakers": 1723917,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -3123,9 +3147,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Walser",
-      "speakers": 11377,
-      "family": "Indo-European",
+      "language_name": "Timne",
+      "speakers": 1722482,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3135,9 +3159,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Wolaytta",
-      "speakers": 1946034,
-      "family": "Ta-Ne-Omotic",
+      "language_name": "Scots",
+      "speakers": 1644028,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3147,21 +3171,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "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": "Warlpiri",
-      "speakers": 2496,
-      "family": "Pama-Nyungan",
+      "language_name": "Lango (Uganda)",
+      "speakers": 1643614,
+      "family": "Nilotic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3171,9 +3183,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Waddar",
-      "speakers": 2386962,
-      "family": "Dravidian",
+      "language_name": "Domari",
+      "speakers": 1613543,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3183,11 +3195,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Wagdi",
-      "speakers": 1989135,
+      "language_name": "Macedonian",
+      "speakers": 1608565,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3195,9 +3207,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Wallisian",
-      "speakers": 9512,
-      "family": "Austronesian",
+      "language_name": "Acoli",
+      "speakers": 1600361,
+      "family": "Nilotic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3207,8 +3219,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ndzwani Comorian",
-      "speakers": 287736,
+      "language_name": "Central-Eastern Niger Fulfulde",
+      "speakers": 1594068,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -3219,11 +3231,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Wolof",
-      "speakers": 11025494,
-      "family": "Atlantic-Congo",
+      "language_name": "Bhili",
+      "speakers": 1591308,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3231,11 +3243,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mewati",
-      "speakers": 6100014,
-      "family": "Indo-European",
+      "language_name": "Pangasinan",
+      "speakers": 1528534,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3243,9 +3255,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Wu Chinese",
-      "speakers": 83641200,
-      "family": "Sino-Tibetan",
+      "language_name": "Kongo",
+      "speakers": 1526700,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3255,9 +3267,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Xavánte",
-      "speakers": 9951,
-      "family": "Nuclear-Macro-Je",
+      "language_name": "Bini",
+      "speakers": 1519599,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3267,9 +3279,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tsakhur",
-      "speakers": 16329,
-      "family": "Nakh-Daghestanian",
+      "language_name": "Maasina Fulfulde",
+      "speakers": 1505612,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3279,9 +3291,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Turkmen",
-      "speakers": 6870838,
-      "family": "Turkic",
+      "language_name": "Manipuri",
+      "speakers": 1476591,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -3291,11 +3303,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sanskrit",
-      "speakers": 15913,
-      "family": "Indo-European",
+      "language_name": "Abron",
+      "speakers": 1467010,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3303,9 +3315,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Koyraboro Senni",
-      "speakers": 664816,
-      "family": "Songhay",
+      "language_name": "Makonde",
+      "speakers": 1463820,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3315,8 +3327,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Samogitian",
-      "speakers": 0,
+      "language_name": "Gheg Albanian",
+      "speakers": 1430250,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -3327,9 +3339,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tachelhit",
-      "speakers": 6187736,
-      "family": "Afro-Asiatic",
+      "language_name": "Venda",
+      "speakers": 1391759,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3339,11 +3351,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Shan",
-      "speakers": 3687984,
-      "family": "Tai-Kadai",
+      "language_name": "Sena",
+      "speakers": 1384517,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3351,8 +3363,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sinhala",
-      "speakers": 15564656,
+      "language_name": "Venetian",
+      "speakers": 1380829,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -3363,9 +3375,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sidamo",
-      "speakers": 3783955,
-      "family": "Afro-Asiatic",
+      "language_name": "Susu",
+      "speakers": 1378014,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3375,8 +3387,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Slovak",
-      "speakers": 6680269,
+      "language_name": "Norwegian Nynorsk",
+      "speakers": 1366860,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -3387,9 +3399,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Saraiki",
-      "speakers": 28020120,
-      "family": "Indo-European",
+      "language_name": "Makhuwa-Meetto",
+      "speakers": 1354419,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3399,20 +3411,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Slovenian",
-      "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": "Lower Silesian",
-      "speakers": 11868,
+      "language_name": "Rajasthani",
+      "speakers": 1326090,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -3423,9 +3423,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Selayar",
-      "speakers": 144194,
-      "family": "Austronesian",
+      "language_name": "Ho",
+      "speakers": 1312829,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3435,11 +3435,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Samoan",
-      "speakers": 252717,
+      "language_name": "Maguindanaon",
+      "speakers": 1310172,
       "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3447,9 +3447,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Southern Sami",
-      "speakers": 296,
-      "family": "Uralic",
+      "language_name": "Mewari",
+      "speakers": 1286307,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3459,9 +3459,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lule Sami",
-      "speakers": 1530,
-      "family": "Uralic",
+      "language_name": "Bulu",
+      "speakers": 1276270,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3471,9 +3471,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Inari Sami",
-      "speakers": 613,
-      "family": "Uralic",
+      "language_name": "Masaaba",
+      "speakers": 1254337,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3483,9 +3483,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Skolt Sami",
-      "speakers": 613,
-      "family": "Uralic",
+      "language_name": "Mundari",
+      "speakers": 1252287,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3495,11 +3495,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Shona",
-      "speakers": 11782503,
-      "family": "Atlantic-Congo",
+      "language_name": "Morisyen",
+      "speakers": 1241433,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3507,11 +3507,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Noon",
-      "speakers": 37767,
-      "family": "Atlantic-Congo",
+      "language_name": "Irish",
+      "speakers": 1237487,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3519,11 +3519,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sango",
-      "speakers": 2935521,
-      "family": "Atlantic-Congo",
+      "language_name": "Fur",
+      "speakers": 1230163,
+      "family": "Furan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3531,9 +3531,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Seri",
-      "speakers": 901,
-      "family": null,
+      "language_name": "Rejang",
+      "speakers": 1228320,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3543,11 +3543,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Somali",
-      "speakers": 16911645,
-      "family": "Afro-Asiatic",
+      "language_name": "Bhilali",
+      "speakers": 1220003,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3555,9 +3555,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sena",
-      "speakers": 1384517,
-      "family": "Atlantic-Congo",
+      "language_name": "Tausug",
+      "speakers": 1200991,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3567,9 +3567,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Laki",
-      "speakers": 645417,
-      "family": "Indo-European",
+      "language_name": "Kʼicheʼ",
+      "speakers": 1200731,
+      "family": "Mayan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3579,9 +3579,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Yakut",
-      "speakers": 453510,
-      "family": "Turkic",
+      "language_name": "Bakhtiari",
+      "speakers": 1188926,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3591,9 +3591,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Samburu",
-      "speakers": 246228,
-      "family": "Nilotic",
+      "language_name": "Kpelle",
+      "speakers": 1186303,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3603,9 +3603,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sasak",
-      "speakers": 2590152,
-      "family": "Austronesian",
+      "language_name": "Cebaara Senoufo",
+      "speakers": 1181687,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3615,11 +3615,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Santali",
-      "speakers": 7293495,
+      "language_name": "Northern Khmer",
+      "speakers": 1172616,
       "family": "Austroasiatic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3627,9 +3627,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Saafi-Saafi",
-      "speakers": 236046,
-      "family": "Atlantic-Congo",
+      "language_name": "West Flemish",
+      "speakers": 1172070,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3639,9 +3639,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Saurashtra",
-      "speakers": 384566,
-      "family": "Indo-European",
+      "language_name": "Soninke",
+      "speakers": 1153651,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3651,9 +3651,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sangu",
-      "speakers": 117106,
-      "family": "Atlantic-Congo",
+      "language_name": "Zaza",
+      "speakers": 1148245,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3663,8 +3663,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sardinian",
-      "speakers": 1060846,
+      "language_name": "Latvian",
+      "speakers": 1147550,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -3675,9 +3675,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sadri",
-      "speakers": 2386962,
-      "family": "Indo-European",
+      "language_name": "Dan",
+      "speakers": 1099244,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3687,11 +3687,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sicilian",
-      "speakers": 511702,
-      "family": "Indo-European",
+      "language_name": "Gorontalo",
+      "speakers": 1094807,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3699,9 +3699,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Scots",
-      "speakers": 1644028,
-      "family": "Indo-European",
+      "language_name": "Tigre",
+      "speakers": 1094616,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3711,9 +3711,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sindhi",
-      "speakers": 40329510,
-      "family": "Indo-European",
+      "language_name": "Basque",
+      "speakers": 1088519,
+      "family": null,
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -3723,8 +3723,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sassarese Sardinian",
-      "speakers": 106085,
+      "language_name": "Hadothi",
+      "speakers": 1087394,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -3735,9 +3735,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Southern Kurdish",
-      "speakers": 3142162,
-      "family": "Indo-European",
+      "language_name": "Kabardian",
+      "speakers": 1070873,
+      "family": "Abkhaz-Adyge",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3747,9 +3747,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern Sami",
-      "speakers": 51530,
-      "family": "Uralic",
+      "language_name": "Khasi",
+      "speakers": 1060872,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3759,11 +3759,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Cebaara Senoufo",
-      "speakers": 1181687,
-      "family": "Atlantic-Congo",
+      "language_name": "Sardinian",
+      "speakers": 1060846,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3771,9 +3771,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Soninke",
-      "speakers": 1153651,
-      "family": "Mande",
+      "language_name": "Lozi",
+      "speakers": 1045596,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3783,9 +3783,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Southern Thai",
-      "speakers": 5518192,
-      "family": "Tai-Kadai",
+      "language_name": "Frafra",
+      "speakers": 1026907,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3795,11 +3795,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tiv",
-      "speakers": 3424448,
-      "family": "Atlantic-Congo",
+      "language_name": "Najdi Arabic",
+      "speakers": 1025205,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3807,11 +3807,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Silesian",
-      "speakers": 497670,
-      "family": "Indo-European",
+      "language_name": "Ronga",
+      "speakers": 1023339,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3819,9 +3819,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Eastern Tamang",
-      "speakers": 130410,
-      "family": "Sino-Tibetan",
+      "language_name": "Southern Luri",
+      "speakers": 1019080,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3831,9 +3831,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tagbanwa",
-      "speakers": 10045,
-      "family": "Austronesian",
+      "language_name": "Newari",
+      "speakers": 1000821,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3843,9 +3843,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tulu",
-      "speakers": 1989135,
-      "family": "Dravidian",
+      "language_name": "Talysh",
+      "speakers": 1000168,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3855,11 +3855,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tai Nüa",
-      "speakers": 264864,
-      "family": "Tai-Kadai",
+      "language_name": "Yiddish",
+      "speakers": 997214,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3867,9 +3867,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Tamang",
-      "speakers": 394263,
-      "family": "Sino-Tibetan",
+      "language_name": "Bena",
+      "speakers": 995398,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3879,9 +3879,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Thulung",
-      "speakers": 36393,
-      "family": "Sino-Tibetan",
+      "language_name": "Shambala",
+      "speakers": 995398,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3891,11 +3891,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Telugu",
-      "speakers": 95478480,
-      "family": "Dravidian",
+      "language_name": "Kachhi",
+      "speakers": 994568,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3903,9 +3903,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Timne",
-      "speakers": 1722482,
-      "family": "Atlantic-Congo",
+      "language_name": "Ngaju",
+      "speakers": 987996,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3915,9 +3915,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Teso",
-      "speakers": 2082973,
-      "family": "Nilotic",
+      "language_name": "Kita Maninkakan",
+      "speakers": 977670,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3925,11 +3925,11 @@
       "language_modeling_chrf": 0.0,
       "translation_bleu": 0.0,
       "translation_chrf": 0.0
-    },
-    {
-      "language_name": "Tetum",
-      "speakers": 816395,
-      "family": "Austronesian",
+    },
+    {
+      "language_name": "Jumli",
+      "speakers": 970493,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3939,11 +3939,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tajik",
-      "speakers": 9644223,
-      "family": "Indo-European",
+      "language_name": "Central Okinawan",
+      "speakers": 966404,
+      "family": "Japonic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3951,11 +3951,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Thai",
-      "speakers": 55181920,
-      "family": "Tai-Kadai",
+      "language_name": "Mon",
+      "speakers": 966114,
+      "family": "Austroasiatic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3963,9 +3963,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dangaura Tharu",
-      "speakers": 606558,
-      "family": "Indo-European",
+      "language_name": "Chimborazo Highland Quichua",
+      "speakers": 963579,
+      "family": "Quechuan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -3975,11 +3975,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kochila Tharu",
-      "speakers": 303279,
-      "family": "Indo-European",
+      "language_name": "Kachin",
+      "speakers": 962032,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -3987,8 +3987,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rana Tharu",
-      "speakers": 363935,
+      "language_name": "Southern Hindko",
+      "speakers": 957354,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -3999,9 +3999,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tigrinya",
-      "speakers": 10145911,
-      "family": "Afro-Asiatic",
+      "language_name": "Limburgish",
+      "speakers": 950422,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -4011,9 +4011,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tigre",
-      "speakers": 1094616,
-      "family": "Afro-Asiatic",
+      "language_name": "Manyika",
+      "speakers": 945510,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4023,11 +4023,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tamil",
-      "speakers": 85616159,
-      "family": "Dravidian",
+      "language_name": "Chechen",
+      "speakers": 935365,
+      "family": "Nakh-Daghestanian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4035,9 +4035,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Syriac",
-      "speakers": 210659,
-      "family": "Afro-Asiatic",
+      "language_name": "Kuanyama",
+      "speakers": 920524,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4047,11 +4047,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Albanian",
-      "speakers": 6791906,
-      "family": "Indo-European",
+      "language_name": "South Ndebele",
+      "speakers": 903418,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4059,11 +4059,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sylheti",
-      "speakers": 8132550,
+      "language_name": "Welsh",
+      "speakers": 884910,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4071,11 +4071,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Serbian",
-      "speakers": 15602410,
-      "family": "Indo-European",
+      "language_name": "Adangme",
+      "speakers": 880206,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4083,11 +4083,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sranan Tongo",
-      "speakers": 414507,
-      "family": "Indo-European",
+      "language_name": "Estonian",
+      "speakers": 878449,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4095,9 +4095,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Serer",
-      "speakers": 1731004,
-      "family": "Atlantic-Congo",
+      "language_name": "Parsi-Dari",
+      "speakers": 864342,
+      "family": "Bookkeeping",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4107,9 +4107,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sirmauri",
-      "speakers": 464132,
-      "family": "Indo-European",
+      "language_name": "Yucateco",
+      "speakers": 861955,
+      "family": "Mayan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4119,11 +4119,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Swati",
-      "speakers": 2212379,
+      "language_name": "Ewondo",
+      "speakers": 860095,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4131,9 +4131,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Saho",
-      "speakers": 218923,
-      "family": "Afro-Asiatic",
+      "language_name": "Komering",
+      "speakers": 854483,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4143,11 +4143,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Southern Sotho",
-      "speakers": 6390567,
+      "language_name": "Tooro",
+      "speakers": 821807,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4155,9 +4155,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Saterland Frisian",
-      "speakers": 962,
-      "family": "Indo-European",
+      "language_name": "Garo",
+      "speakers": 821563,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4167,11 +4167,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sundanese",
-      "speakers": 32043120,
-      "family": "Austronesian",
+      "language_name": "Ga",
+      "speakers": 821526,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4179,8 +4179,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sukuma",
-      "speakers": 5094094,
+      "language_name": "Mbunga",
+      "speakers": 819739,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -4191,9 +4191,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Susu",
-      "speakers": 1378014,
-      "family": "Mande",
+      "language_name": "Tetum",
+      "speakers": 816395,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4203,11 +4203,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Swedish",
-      "speakers": 12932871,
-      "family": "Indo-European",
+      "language_name": "Iban",
+      "speakers": 816302,
+      "family": "Bookkeeping",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4215,11 +4215,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Swahili",
-      "speakers": 171610296,
+      "language_name": "Embu",
+      "speakers": 802918,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4227,9 +4227,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Comorian",
-      "speakers": 170720,
-      "family": "Atlantic-Congo",
+      "language_name": "Swabian",
+      "speakers": 801597,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4239,9 +4239,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Swabian",
-      "speakers": 801597,
-      "family": "Indo-European",
+      "language_name": "Hmong Njua",
+      "speakers": 781687,
+      "family": "Hmong-Mien",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4251,9 +4251,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Shekhawati",
-      "speakers": 3713052,
-      "family": "Indo-European",
+      "language_name": "Kalanga",
+      "speakers": 770954,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4263,9 +4263,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sangir",
-      "speakers": 245664,
-      "family": "Austronesian",
+      "language_name": "Dotyali",
+      "speakers": 758198,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4275,9 +4275,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lakota",
-      "speakers": 8316,
-      "family": "Siouan",
+      "language_name": "Picard",
+      "speakers": 746330,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4287,9 +4287,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lillooet",
-      "speakers": 528,
-      "family": "Salishan",
+      "language_name": "Western Frisian",
+      "speakers": 743057,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4299,9 +4299,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lampung Api",
-      "speakers": 1842479,
-      "family": "Austronesian",
+      "language_name": "Chakma",
+      "speakers": 729137,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4311,9 +4311,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Cherokee",
-      "speakers": 25613,
-      "family": "Iroquoian",
+      "language_name": "Yao",
+      "speakers": 722357,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4323,8 +4323,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Cham",
-      "speakers": 270832,
+      "language_name": "Uab Meto",
+      "speakers": 720970,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
@@ -4335,8 +4335,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Eastern Cham",
-      "speakers": 87862,
+      "language_name": "Capiznon",
+      "speakers": 720595,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
@@ -4347,11 +4347,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central Kurdish",
-      "speakers": 11086549,
-      "family": "Indo-European",
+      "language_name": "Asu",
+      "speakers": 702634,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4359,9 +4359,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chilcotin",
-      "speakers": 867,
-      "family": "Athabaskan-Eyak-Tlingit",
+      "language_name": "Tai Dam",
+      "speakers": 681177,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4371,8 +4371,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Corsican",
-      "speakers": 162836,
+      "language_name": "Walloon",
+      "speakers": 679801,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -4383,9 +4383,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Capiznon",
-      "speakers": 720595,
-      "family": "Austronesian",
+      "language_name": "Koyraboro Senni",
+      "speakers": 664816,
+      "family": "Songhay",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4395,9 +4395,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Cree",
-      "speakers": 9047,
-      "family": "Algic",
+      "language_name": "Bagheli",
+      "speakers": 654424,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4407,11 +4407,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Michif",
-      "speakers": 678,
-      "family": "Algic",
+      "language_name": "Asturian",
+      "speakers": 650205,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4419,11 +4419,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Crimean Tatar",
-      "speakers": 245968,
-      "family": "Turkic",
+      "language_name": "Laki",
+      "speakers": 645417,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4431,9 +4431,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Plains Cree",
-      "speakers": 4146,
-      "family": "Algic",
+      "language_name": "Munda",
+      "speakers": 636523,
+      "family": "Bookkeeping",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4443,9 +4443,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern East Cree",
-      "speakers": 377,
-      "family": "Algic",
+      "language_name": "Lisu",
+      "speakers": 627309,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4455,11 +4455,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Seselwa Creole French",
-      "speakers": 94061,
-      "family": "Indo-European",
+      "language_name": "N’Ko",
+      "speakers": 626370,
+      "family": "Artificial Language",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4467,11 +4467,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Czech",
-      "speakers": 13045532,
+      "language_name": "Gronings",
+      "speakers": 622094,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4479,9 +4479,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kashubian",
-      "speakers": 49767,
-      "family": "Indo-European",
+      "language_name": "Mongo",
+      "speakers": 620858,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4491,9 +4491,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Swampy Cree",
-      "speakers": 1809,
-      "family": "Algic",
+      "language_name": "Dangaura Tharu",
+      "speakers": 606558,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4503,8 +4503,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Church Slavic",
-      "speakers": 0,
+      "language_name": "Neapolitan",
+      "speakers": 605306,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -4515,9 +4515,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chuvash",
-      "speakers": 1842386,
-      "family": "Turkic",
+      "language_name": "Nuer",
+      "speakers": 591427,
+      "family": "Nilotic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -4527,9 +4527,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chickasaw",
-      "speakers": 0,
-      "family": "Muskogean",
+      "language_name": "Breton",
+      "speakers": 563140,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4539,9 +4539,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chipewyan",
-      "speakers": 12816,
-      "family": "Athabaskan-Eyak-Tlingit",
+      "language_name": "Pijin",
+      "speakers": 561780,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4551,11 +4551,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Danish",
-      "speakers": 7072056,
+      "language_name": "Réunion Creole French",
+      "speakers": 559185,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4563,9 +4563,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Choctaw",
-      "speakers": 10977,
-      "family": "Muskogean",
+      "language_name": "Avaric",
+      "speakers": 552716,
+      "family": "Nakh-Daghestanian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4575,9 +4575,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bushi",
-      "speakers": 44620,
-      "family": "Austronesian",
+      "language_name": "Ndonga",
+      "speakers": 552315,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4587,11 +4587,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Buginese",
-      "speakers": 4298211,
-      "family": "Austronesian",
+      "language_name": "Ossetic",
+      "speakers": 541444,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4599,9 +4599,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bulu",
-      "speakers": 1276270,
-      "family": "Atlantic-Congo",
+      "language_name": "Udmurt",
+      "speakers": 538544,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4611,11 +4611,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bube",
-      "speakers": 66058,
-      "family": "Atlantic-Congo",
+      "language_name": "Ligurian",
+      "speakers": 536663,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4623,11 +4623,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Blin",
-      "speakers": 79056,
-      "family": "Afro-Asiatic",
+      "language_name": "Kabuverdianu",
+      "speakers": 530762,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4635,9 +4635,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Medumba",
-      "speakers": 305195,
-      "family": "Atlantic-Congo",
+      "language_name": "Rusyn",
+      "speakers": 527075,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4647,11 +4647,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Jenaama Bozo",
-      "speakers": 166204,
-      "family": "Mande",
+      "language_name": "Mari",
+      "speakers": 524371,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4659,8 +4659,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Catalan",
-      "speakers": 8679139,
+      "language_name": "Sicilian",
+      "speakers": 511702,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -4671,9 +4671,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Caddo",
-      "speakers": 0,
-      "family": "Caddoan",
+      "language_name": "Langi",
+      "speakers": 509409,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4683,9 +4683,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Atsam",
-      "speakers": 44946,
-      "family": "Atlantic-Congo",
+      "language_name": "Balti",
+      "speakers": 502520,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4695,9 +4695,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chakma",
-      "speakers": 729137,
-      "family": "Indo-European",
+      "language_name": "Eastern Huasteca Nahuatl",
+      "speakers": 501735,
+      "family": "Uto-Aztecan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4707,9 +4707,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chechen",
-      "speakers": 935365,
-      "family": "Nakh-Daghestanian",
+      "language_name": "Western Huasteca Nahuatl",
+      "speakers": 501735,
+      "family": "Uto-Aztecan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4719,9 +4719,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Cebuano",
-      "speakers": 26203440,
-      "family": "Austronesian",
+      "language_name": "Silesian",
+      "speakers": 497670,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -4731,9 +4731,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chiga",
-      "speakers": 2335662,
-      "family": "Atlantic-Congo",
+      "language_name": "Kara-Kalpak",
+      "speakers": 489046,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4743,9 +4743,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chamorro",
-      "speakers": 46325,
-      "family": "Austronesian",
+      "language_name": "Gujari",
+      "speakers": 467002,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4755,9 +4755,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chuukese",
-      "speakers": 30731,
-      "family": "Austronesian",
+      "language_name": "Sirmauri",
+      "speakers": 464132,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4767,9 +4767,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mari",
-      "speakers": 524371,
-      "family": "Uralic",
+      "language_name": "Maltese",
+      "speakers": 457267,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -4779,11 +4779,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Welsh",
-      "speakers": 884910,
-      "family": "Indo-European",
+      "language_name": "Bantawa",
+      "speakers": 454918,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4791,9 +4791,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dakota",
-      "speakers": 20832,
-      "family": "Siouan",
+      "language_name": "Yakut",
+      "speakers": 453510,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4803,9 +4803,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bateri",
-      "speakers": 78843,
-      "family": "Indo-European",
+      "language_name": "Masalit",
+      "speakers": 451060,
+      "family": "Maban",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4815,8 +4815,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Efik",
-      "speakers": 2996392,
+      "language_name": "Jju",
+      "speakers": 449459,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -4827,11 +4827,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Greek",
-      "speakers": 12292242,
-      "family": "Indo-European",
+      "language_name": "Adyghe",
+      "speakers": 444583,
+      "family": "Abkhaz-Adyge",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4839,11 +4839,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Esperanto",
-      "speakers": 301,
-      "family": "Artificial Language",
+      "language_name": "Yemba",
+      "speakers": 443920,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4851,9 +4851,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central Yupik",
-      "speakers": 20956,
-      "family": "Eskimo-Aleut",
+      "language_name": "Mingrelian",
+      "speakers": 439670,
+      "family": "Kartvelian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4863,32 +4863,20 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Estonian",
-      "speakers": 878449,
+      "language_name": "Erzya",
+      "speakers": 439338,
       "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": "Basque",
-      "speakers": 1088519,
-      "family": null,
-      "average": 0.0,
-      "in_benchmark": true,
+      "in_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",
-      "speakers": 860095,
+    {
+      "language_name": "Taita",
+      "speakers": 438929,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -4899,9 +4887,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Extremaduran",
-      "speakers": 245077,
-      "family": "Indo-European",
+      "language_name": "Central Mazahua",
+      "speakers": 437410,
+      "family": "Otomanguean",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4911,11 +4899,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Persian",
-      "speakers": 84710459,
-      "family": "Indo-European",
+      "language_name": "Vunjo",
+      "speakers": 433291,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4923,8 +4911,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Fang",
-      "speakers": 426451,
+      "language_name": "Rombo",
+      "speakers": 433291,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -4935,9 +4923,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "West Albay Bikol",
-      "speakers": 2511163,
-      "family": "Austronesian",
+      "language_name": "Machame",
+      "speakers": 433291,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4947,9 +4935,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Fula",
-      "speakers": 7788904,
-      "family": "Atlantic-Congo",
+      "language_name": "Khmu",
+      "speakers": 431949,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4959,8 +4947,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Maasina Fulfulde",
-      "speakers": 1505612,
+      "language_name": "Fang",
+      "speakers": 426451,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -4971,11 +4959,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Finnish",
-      "speakers": 5736842,
-      "family": "Uralic",
+      "language_name": "Kinaray-a",
+      "speakers": 425806,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -4983,9 +4971,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nobiin",
-      "speakers": 378161,
-      "family": "Nubian",
+      "language_name": "Kuy",
+      "speakers": 421207,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -4995,9 +4983,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Filipino",
-      "speakers": 67471096,
-      "family": "Austronesian",
+      "language_name": "Luxembourgish",
+      "speakers": 421015,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -5007,9 +4995,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tornedalen Finnish",
-      "speakers": 56114,
-      "family": "Uralic",
+      "language_name": "Sranan Tongo",
+      "speakers": 414507,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5019,11 +5007,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Fijian",
-      "speakers": 365030,
-      "family": "Austronesian",
+      "language_name": "Jola-Fonyi",
+      "speakers": 409146,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5031,9 +5019,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Emilian",
-      "speakers": 31201,
-      "family": "Indo-European",
+      "language_name": "Western Tamang",
+      "speakers": 394263,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5043,11 +5031,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ewe",
-      "speakers": 4690857,
-      "family": "Atlantic-Congo",
+      "language_name": "Pontic",
+      "speakers": 392463,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5055,9 +5043,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dargwa",
-      "speakers": 368477,
-      "family": "Nakh-Daghestanian",
+      "language_name": "Negeri Sembilan Malay",
+      "speakers": 391825,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5067,8 +5055,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Embu",
-      "speakers": 802918,
+      "language_name": "Ngiemboon",
+      "speakers": 388430,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -5079,8 +5067,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Taita",
-      "speakers": 438929,
+      "language_name": "Ghomala",
+      "speakers": 388430,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -5091,8 +5079,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Deccan",
-      "speakers": 13128291,
+      "language_name": "Divehi",
+      "speakers": 388044,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -5103,11 +5091,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "German",
-      "speakers": 136350226,
+      "language_name": "Saurashtra",
+      "speakers": 384566,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5115,9 +5103,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Slave",
-      "speakers": 2299,
-      "family": "Athabaskan-Eyak-Tlingit",
+      "language_name": "Fiji Hindi",
+      "speakers": 383749,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5127,9 +5115,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dogrib",
-      "speakers": 2111,
-      "family": "Athabaskan-Eyak-Tlingit",
+      "language_name": "Nobiin",
+      "speakers": 378161,
+      "family": "Nubian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5139,9 +5127,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Zarma",
-      "speakers": 3871308,
-      "family": "Songhay",
+      "language_name": "Balkan Gagauz Turkish",
+      "speakers": 377280,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5151,9 +5139,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dan",
-      "speakers": 1099244,
-      "family": "Mande",
+      "language_name": "Kerinci",
+      "speakers": 373836,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5163,11 +5151,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dogri",
-      "speakers": 2652180,
+      "language_name": "Parkari Koli",
+      "speakers": 373602,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5175,11 +5163,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lower Sorbian",
-      "speakers": 6974,
-      "family": "Indo-European",
+      "language_name": "Dzongkha",
+      "speakers": 370341,
+      "family": "Bookkeeping",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5187,9 +5175,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tomo Kan Dogon",
-      "speakers": 215087,
-      "family": "Dogon",
+      "language_name": "Pökoot",
+      "speakers": 369343,
+      "family": "Nilotic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5199,9 +5187,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central Dusun",
-      "speakers": 182852,
-      "family": "Austronesian",
+      "language_name": "Dargwa",
+      "speakers": 368477,
+      "family": "Nakh-Daghestanian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5211,9 +5199,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dotyali",
-      "speakers": 758198,
-      "family": "Indo-European",
+      "language_name": "Limbu",
+      "speakers": 368085,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5223,11 +5211,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Duala",
-      "speakers": 133176,
-      "family": "Atlantic-Congo",
+      "language_name": "Fijian",
+      "speakers": 365030,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5235,8 +5223,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Divehi",
-      "speakers": 388044,
+      "language_name": "Rana Tharu",
+      "speakers": 363935,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -5247,8 +5235,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Jola-Fonyi",
-      "speakers": 409146,
+      "language_name": "Kom",
+      "speakers": 360685,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -5259,9 +5247,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dyula",
-      "speakers": 6667328,
-      "family": "Mande",
+      "language_name": "Icelandic",
+      "speakers": 350734,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -5271,11 +5259,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Dzongkha",
-      "speakers": 370341,
-      "family": "Bookkeeping",
+      "language_name": "Khowar",
+      "speakers": 350252,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5283,9 +5271,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Buriat",
-      "speakers": 311788,
-      "family": "Mongolic-Khitan",
+      "language_name": "Tolaki",
+      "speakers": 347134,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5295,9 +5283,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Rinconada Bikol",
-      "speakers": 305707,
-      "family": "Austronesian",
+      "language_name": "Naxi",
+      "speakers": 334565,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5307,11 +5295,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Fon",
-      "speakers": 3216150,
-      "family": "Atlantic-Congo",
+      "language_name": "Eastern Magar",
+      "speakers": 333607,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5319,9 +5307,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Algerian Arabic",
-      "speakers": 35667507,
-      "family": "Afro-Asiatic",
+      "language_name": "Bamun",
+      "speakers": 332940,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5331,11 +5319,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Moroccan Arabic",
-      "speakers": 30938679,
-      "family": "Afro-Asiatic",
+      "language_name": "Basaa",
+      "speakers": 332940,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5343,11 +5331,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Egyptian Arabic",
-      "speakers": 66639360,
-      "family": "Afro-Asiatic",
+      "language_name": "Koyra Chiini",
+      "speakers": 332408,
+      "family": "Songhay",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5355,11 +5343,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Assamese",
-      "speakers": 17239170,
+      "language_name": "Indus Kohistani",
+      "speakers": 326901,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5367,8 +5355,20 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Asu",
-      "speakers": 702634,
+      "language_name": "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",
+      "speakers": 313124,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -5379,11 +5379,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Asturian",
-      "speakers": 650205,
-      "family": "Indo-European",
+      "language_name": "Buriat",
+      "speakers": 311788,
+      "family": "Mongolic-Khitan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5391,9 +5391,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Atikamekw",
-      "speakers": 6408,
-      "family": "Algic",
+      "language_name": "Rinconada Bikol",
+      "speakers": 305707,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5403,9 +5403,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Avaric",
-      "speakers": 552716,
-      "family": "Nakh-Daghestanian",
+      "language_name": "Medumba",
+      "speakers": 305195,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5415,11 +5415,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Awadhi",
-      "speakers": 25862924,
-      "family": "Indo-European",
+      "language_name": "Ao Naga",
+      "speakers": 305001,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5427,11 +5427,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Aymara",
-      "speakers": 2838620,
-      "family": "Aymaran",
+      "language_name": "Badaga",
+      "speakers": 305001,
+      "family": "Dravidian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5439,11 +5439,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Azerbaijani",
-      "speakers": 32446682,
-      "family": "Turkic",
+      "language_name": "Kochila Tharu",
+      "speakers": 303279,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5451,11 +5451,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bashkir",
-      "speakers": 1842386,
-      "family": "Turkic",
+      "language_name": "Moksha",
+      "speakers": 297616,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5463,9 +5463,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Baluchi",
-      "speakers": 8227887,
-      "family": "Indo-European",
+      "language_name": "Tae'",
+      "speakers": 293729,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5475,11 +5475,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Balinese",
-      "speakers": 4806468,
-      "family": "Austronesian",
+      "language_name": "Nzima",
+      "speakers": 293402,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5487,9 +5487,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bantawa",
-      "speakers": 454918,
-      "family": "Sino-Tibetan",
+      "language_name": "Nama",
+      "speakers": 289308,
+      "family": "Khoe-Kwadi",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5499,9 +5499,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bavarian",
-      "speakers": 22043627,
-      "family": "Indo-European",
+      "language_name": "Ndzwani Comorian",
+      "speakers": 287736,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5511,9 +5511,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Basaa",
-      "speakers": 332940,
-      "family": "Atlantic-Congo",
+      "language_name": "Kumyk",
+      "speakers": 283444,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5523,8 +5523,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bamun",
-      "speakers": 332940,
+      "language_name": "Mundang",
+      "speakers": 277450,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -5535,11 +5535,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Najdi Arabic",
-      "speakers": 1025205,
-      "family": "Afro-Asiatic",
+      "language_name": "Mapuche",
+      "speakers": 272802,
+      "family": "Araucanian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5547,9 +5547,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Araona",
-      "speakers": 105,
-      "family": "Pano-Tacanan",
+      "language_name": "Western Cham",
+      "speakers": 270832,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5559,9 +5559,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ghomala",
-      "speakers": 388430,
-      "family": "Atlantic-Congo",
+      "language_name": "Bislama",
+      "speakers": 268500,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5571,9 +5571,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mapuche",
-      "speakers": 272802,
-      "family": "Araucanian",
+      "language_name": "Lü",
+      "speakers": 264864,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5583,9 +5583,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Abron",
-      "speakers": 1467010,
-      "family": "Atlantic-Congo",
+      "language_name": "Tai Nüa",
+      "speakers": 264864,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5595,11 +5595,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Achinese",
-      "speakers": 3738364,
-      "family": "Austronesian",
+      "language_name": "Wadiyara Koli",
+      "speakers": 256851,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5607,9 +5607,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Acoli",
-      "speakers": 1600361,
-      "family": "Nilotic",
+      "language_name": "Kachi Koli",
+      "speakers": 256851,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5619,9 +5619,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Adangme",
-      "speakers": 880206,
-      "family": "Atlantic-Congo",
+      "language_name": "Lezghian",
+      "speakers": 255100,
+      "family": "Nakh-Daghestanian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5631,9 +5631,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Adyghe",
-      "speakers": 444583,
-      "family": "Abkhaz-Adyge",
+      "language_name": "Komi",
+      "speakers": 255100,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5643,9 +5643,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tunisian Arabic",
-      "speakers": 10549080,
-      "family": "Afro-Asiatic",
+      "language_name": "Samoan",
+      "speakers": 252717,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -5655,11 +5655,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Afrikaans",
-      "speakers": 9318845,
-      "family": "Indo-European",
+      "language_name": "Western Magar",
+      "speakers": 251722,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5667,9 +5667,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Aghem",
-      "speakers": 38843,
-      "family": "Atlantic-Congo",
+      "language_name": "Samburu",
+      "speakers": 246228,
+      "family": "Nilotic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5679,11 +5679,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Akan",
-      "speakers": 11442678,
-      "family": "Atlantic-Congo",
+      "language_name": "Crimean Tatar",
+      "speakers": 245968,
+      "family": "Turkic",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5691,9 +5691,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gheg Albanian",
-      "speakers": 1430250,
-      "family": "Indo-European",
+      "language_name": "Mandar",
+      "speakers": 245664,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5703,9 +5703,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Southern Altai",
-      "speakers": 19841,
-      "family": "Turkic",
+      "language_name": "Sangir",
+      "speakers": 245664,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5715,11 +5715,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Amharic",
-      "speakers": 35728475,
-      "family": "Afro-Asiatic",
+      "language_name": "Extremaduran",
+      "speakers": 245077,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5727,9 +5727,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Amo",
-      "speakers": 18620,
-      "family": "Atlantic-Congo",
+      "language_name": "Central Huasteca Nahuatl",
+      "speakers": 244435,
+      "family": "Uto-Aztecan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5739,8 +5739,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Aragonese",
-      "speakers": 26008,
+      "language_name": "Zeelandic",
+      "speakers": 241926,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -5751,9 +5751,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Obolo",
-      "speakers": 0,
-      "family": "Atlantic-Congo",
+      "language_name": "Karachay-Balkar",
+      "speakers": 240927,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5763,9 +5763,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Uab Meto",
-      "speakers": 720970,
-      "family": "Austronesian",
+      "language_name": "Colognian",
+      "speakers": 240479,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5775,9 +5775,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "North Levantine Arabic",
-      "speakers": 39031474,
-      "family": "Afro-Asiatic",
+      "language_name": "Herero",
+      "speakers": 239336,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5787,9 +5787,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Batak Toba",
-      "speakers": 2456639,
-      "family": "Austronesian",
+      "language_name": "Saafi-Saafi",
+      "speakers": 236046,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5799,9 +5799,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Baoulé",
-      "speakers": 3022921,
-      "family": "Atlantic-Congo",
+      "language_name": "Zaghawa",
+      "speakers": 232364,
+      "family": "Saharan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5811,9 +5811,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Akoose",
-      "speakers": 149823,
-      "family": "Atlantic-Congo",
+      "language_name": "Ingush",
+      "speakers": 226755,
+      "family": "Nakh-Daghestanian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5823,11 +5823,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Banjar",
-      "speakers": 4010288,
-      "family": "Austronesian",
+      "language_name": "Saho",
+      "speakers": 218923,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5835,9 +5835,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kom",
-      "speakers": 360685,
-      "family": "Atlantic-Congo",
+      "language_name": "Tomo Kan Dogon",
+      "speakers": 215087,
+      "family": "Dogon",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5847,11 +5847,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Buhid",
-      "speakers": 7970,
-      "family": "Austronesian",
+      "language_name": "Papiamento",
+      "speakers": 211640,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5859,9 +5859,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Siksiká",
-      "speakers": 4900,
-      "family": "Algic",
+      "language_name": "Syriac",
+      "speakers": 210659,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5871,8 +5871,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Anii",
-      "speakers": 51507,
+      "language_name": "Nyasa Tonga",
+      "speakers": 207727,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -5883,9 +5883,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tai Dam",
-      "speakers": 681177,
-      "family": "Tai-Kadai",
+      "language_name": "Mafa",
+      "speakers": 205313,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5895,11 +5895,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bambara",
-      "speakers": 9385632,
-      "family": "Mande",
+      "language_name": "Punu",
+      "speakers": 200782,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5907,8 +5907,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bomu",
-      "speakers": 168159,
+      "language_name": "Tyap",
+      "speakers": 199046,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -5919,11 +5919,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tibetan",
-      "speakers": 3006697,
-      "family": "Sino-Tibetan",
+      "language_name": "Xaasongaxango",
+      "speakers": 195534,
+      "family": "Mande",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -5931,9 +5931,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bishnupriya",
-      "speakers": 90174,
-      "family": "Indo-European",
+      "language_name": "Tuvinian",
+      "speakers": 184239,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5943,9 +5943,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bakhtiari",
-      "speakers": 1188926,
-      "family": "Indo-European",
+      "language_name": "Central Dusun",
+      "speakers": 182852,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5955,9 +5955,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Koro Wachi",
-      "speakers": 46718,
-      "family": "Atlantic-Congo",
+      "language_name": "Riang (India)",
+      "speakers": 172392,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5967,9 +5967,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Breton",
-      "speakers": 563140,
-      "family": "Indo-European",
+      "language_name": "Comorian",
+      "speakers": 170720,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5979,9 +5979,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Braj",
-      "speakers": 54370,
-      "family": "Indo-European",
+      "language_name": "Bomu",
+      "speakers": 168159,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -5991,11 +5991,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Brahui",
-      "speakers": 3035513,
-      "family": "Dravidian",
+      "language_name": "Latgalian",
+      "speakers": 167429,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6003,11 +6003,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bodo",
-      "speakers": 1856526,
-      "family": "Sino-Tibetan",
+      "language_name": "Navajo",
+      "speakers": 166320,
+      "family": "Athabaskan-Eyak-Tlingit",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6015,11 +6015,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bosnian",
-      "speakers": 7594468,
-      "family": "Indo-European",
+      "language_name": "Jenaama Bozo",
+      "speakers": 166204,
+      "family": "Mande",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6027,9 +6027,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bassari",
-      "speakers": 15264,
-      "family": "Atlantic-Congo",
+      "language_name": "Corsican",
+      "speakers": 162836,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6039,8 +6039,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Balanta-Ganja",
-      "speakers": 95992,
+      "language_name": "Bafut",
+      "speakers": 158146,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -6051,9 +6051,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kanauji",
-      "speakers": 7426104,
-      "family": "Indo-European",
+      "language_name": "Sherpa",
+      "speakers": 157705,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6063,11 +6063,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Belarusian",
-      "speakers": 10064517,
+      "language_name": "Kirmanjki",
+      "speakers": 155833,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6075,9 +6075,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bini",
-      "speakers": 1519599,
-      "family": "Atlantic-Congo",
+      "language_name": "Hiri Motu",
+      "speakers": 152449,
+      "family": "Pidgin",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6087,9 +6087,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Beja",
-      "speakers": 2460326,
-      "family": "Afro-Asiatic",
+      "language_name": "Kako",
+      "speakers": 149823,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6099,11 +6099,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bemba",
-      "speakers": 5402246,
+      "language_name": "Akoose",
+      "speakers": 149823,
       "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6111,8 +6111,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Betawi",
-      "speakers": 5607546,
+      "language_name": "Selayar",
+      "speakers": 144194,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
@@ -6123,11 +6123,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bena",
-      "speakers": 995398,
-      "family": "Atlantic-Congo",
+      "language_name": "Māori",
+      "speakers": 137913,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6135,9 +6135,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bafut",
-      "speakers": 158146,
-      "family": "Atlantic-Congo",
+      "language_name": "Rajbanshi",
+      "speakers": 133443,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6147,9 +6147,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Badaga",
-      "speakers": 305001,
-      "family": "Dravidian",
+      "language_name": "Duala",
+      "speakers": 133176,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6159,9 +6159,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Balti",
-      "speakers": 502520,
-      "family": "Sino-Tibetan",
+      "language_name": "Wayuu",
+      "speakers": 132529,
+      "family": "Arawakan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6171,9 +6171,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bagheli",
-      "speakers": 654424,
-      "family": "Indo-European",
+      "language_name": "Vai",
+      "speakers": 131906,
+      "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6183,11 +6183,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bulgarian",
-      "speakers": 7878315,
-      "family": "Indo-European",
+      "language_name": "Eastern Tamang",
+      "speakers": 130410,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6195,9 +6195,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Haryanvi",
-      "speakers": 15913080,
-      "family": "Indo-European",
+      "language_name": "Metaʼ",
+      "speakers": 130401,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6207,8 +6207,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Balochi",
-      "speakers": 2037382,
+      "language_name": "Pennsylvania German",
+      "speakers": 129729,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -6219,9 +6219,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Balkan Gagauz Turkish",
-      "speakers": 377280,
-      "family": "Turkic",
+      "language_name": "Rwa",
+      "speakers": 128816,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6231,8 +6231,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bhili",
-      "speakers": 1591308,
+      "language_name": "Torwali",
+      "speakers": 123756,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -6243,9 +6243,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bhilali",
-      "speakers": 1220003,
-      "family": "Indo-European",
+      "language_name": "Mandjak",
+      "speakers": 121170,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6255,11 +6255,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bhojpuri",
-      "speakers": 32934797,
-      "family": "Indo-European",
+      "language_name": "Tshangla",
+      "speakers": 117348,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6267,9 +6267,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bislama",
-      "speakers": 268500,
-      "family": "Indo-European",
+      "language_name": "Sangu",
+      "speakers": 117106,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6279,9 +6279,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bikol",
-      "speakers": 3275430,
-      "family": "Austronesian",
+      "language_name": "Karelian",
+      "speakers": 116212,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6291,11 +6291,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Faroese",
-      "speakers": 71351,
+      "language_name": "Ladino",
+      "speakers": 112781,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6303,9 +6303,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Cajun French",
-      "speakers": 27942,
-      "family": "Indo-European",
+      "language_name": "Ifè",
+      "speakers": 111910,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6315,9 +6315,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lisu",
-      "speakers": 627309,
-      "family": "Sino-Tibetan",
+      "language_name": "Gagauz",
+      "speakers": 111028,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6327,9 +6327,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Komering",
-      "speakers": 854483,
-      "family": "Austronesian",
+      "language_name": "Lak",
+      "speakers": 110543,
+      "family": "Nakh-Daghestanian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6339,9 +6339,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Khasi",
-      "speakers": 1060872,
-      "family": "Austroasiatic",
+      "language_name": "Sassarese Sardinian",
+      "speakers": 106085,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6351,9 +6351,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lü",
-      "speakers": 264864,
-      "family": "Tai-Kadai",
+      "language_name": "Tongan",
+      "speakers": 100790,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6363,9 +6363,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Khandesi",
-      "speakers": 1989135,
-      "family": "Indo-European",
+      "language_name": "Balanta-Ganja",
+      "speakers": 95992,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6375,9 +6375,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Koyra Chiini",
-      "speakers": 332408,
-      "family": "Songhay",
+      "language_name": "Ngomba",
+      "speakers": 94333,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6387,9 +6387,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Khamti",
-      "speakers": 13527,
-      "family": "Tai-Kadai",
+      "language_name": "Seselwa Creole French",
+      "speakers": 94061,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6399,9 +6399,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Khowar",
-      "speakers": 350252,
-      "family": "Indo-European",
+      "language_name": "Abkhazian",
+      "speakers": 91953,
+      "family": "Abkhaz-Adyge",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6411,11 +6411,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kikuyu",
-      "speakers": 9099743,
-      "family": "Atlantic-Congo",
+      "language_name": "Tahitian",
+      "speakers": 91488,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6423,9 +6423,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kirmanjki",
-      "speakers": 155833,
-      "family": "Indo-European",
+      "language_name": "Inuktitut",
+      "speakers": 90466,
+      "family": "Eskimo-Aleut",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6435,9 +6435,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kuanyama",
-      "speakers": 920524,
-      "family": "Atlantic-Congo",
+      "language_name": "Plautdietsch",
+      "speakers": 90466,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6447,9 +6447,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Khmu",
-      "speakers": 431949,
-      "family": "Austroasiatic",
+      "language_name": "Bishnupriya",
+      "speakers": 90174,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6459,11 +6459,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kazakh",
-      "speakers": 13637392,
-      "family": "Turkic",
+      "language_name": "Bafia",
+      "speakers": 88784,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6471,9 +6471,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kako",
-      "speakers": 149823,
-      "family": "Atlantic-Congo",
+      "language_name": "Gurung",
+      "speakers": 87951,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6483,9 +6483,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kalaallisut",
-      "speakers": 55440,
-      "family": "Eskimo-Aleut",
+      "language_name": "Eastern Cham",
+      "speakers": 87862,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6495,9 +6495,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kalenjin",
-      "speakers": 4068120,
-      "family": "Nilotic",
+      "language_name": "Western Lawa",
+      "speakers": 87751,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6507,11 +6507,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Khmer",
-      "speakers": 15065030,
-      "family": "Austroasiatic",
+      "language_name": "Mankanya",
+      "speakers": 83151,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6519,11 +6519,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kimbundu",
-      "speakers": 8130575,
-      "family": "Atlantic-Congo",
+      "language_name": "Lepcha",
+      "speakers": 79743,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6531,11 +6531,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kannada",
-      "speakers": 49065330,
-      "family": "Dravidian",
+      "language_name": "Blin",
+      "speakers": 79056,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6543,9 +6543,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kaingang",
-      "speakers": 50812,
-      "family": "Nuclear-Macro-Je",
+      "language_name": "Bateri",
+      "speakers": 78843,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6555,9 +6555,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kongo",
-      "speakers": 1526700,
-      "family": "Atlantic-Congo",
+      "language_name": "Kathoriya Tharu",
+      "speakers": 72787,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6567,9 +6567,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Korean",
-      "speakers": 78357046,
-      "family": "Koreanic",
+      "language_name": "Scottish Gaelic",
+      "speakers": 72337,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
       "NaN": 0.0,
@@ -6579,11 +6579,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kumaoni",
-      "speakers": 2917398,
+      "language_name": "Faroese",
+      "speakers": 71351,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6591,11 +6591,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Javanese",
-      "speakers": 91180665,
-      "family": "Austronesian",
+      "language_name": "Võro",
+      "speakers": 70031,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6603,11 +6603,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Georgian",
-      "speakers": 3543646,
-      "family": "Kartvelian",
+      "language_name": "Kenyang",
+      "speakers": 69362,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6615,9 +6615,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kara-Kalpak",
-      "speakers": 489046,
-      "family": "Turkic",
+      "language_name": "Gilbertese",
+      "speakers": 67078,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6627,11 +6627,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kabyle",
-      "speakers": 3351886,
-      "family": "Afro-Asiatic",
+      "language_name": "Bube",
+      "speakers": 66058,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6639,11 +6639,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kachin",
-      "speakers": 962032,
-      "family": "Sino-Tibetan",
+      "language_name": "Mentawai",
+      "speakers": 64086,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6651,9 +6651,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Jju",
-      "speakers": 449459,
-      "family": "Atlantic-Congo",
+      "language_name": "Arpitan",
+      "speakers": 63777,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6663,11 +6663,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kamba",
-      "speakers": 4068120,
-      "family": "Atlantic-Congo",
+      "language_name": "Komi-Permyak",
+      "speakers": 63775,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6675,8 +6675,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Xaasongaxango",
-      "speakers": 195534,
+      "language_name": "Koro",
+      "speakers": 63207,
       "family": "Mande",
       "average": 0.0,
       "in_benchmark": false,
@@ -6687,9 +6687,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kabardian",
-      "speakers": 1070873,
-      "family": "Abkhaz-Adyge",
+      "language_name": "Marshallese",
+      "speakers": 56879,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6699,9 +6699,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tyap",
-      "speakers": 199046,
-      "family": "Atlantic-Congo",
+      "language_name": "Tornedalen Finnish",
+      "speakers": 56114,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6711,9 +6711,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kalanga",
-      "speakers": 770954,
-      "family": "Atlantic-Congo",
+      "language_name": "Kalaallisut",
+      "speakers": 55440,
+      "family": "Eskimo-Aleut",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6723,9 +6723,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Makonde",
-      "speakers": 1463820,
-      "family": "Atlantic-Congo",
+      "language_name": "Braj",
+      "speakers": 54370,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6735,9 +6735,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kuy",
-      "speakers": 421207,
-      "family": "Austroasiatic",
+      "language_name": "Guianese Creole French",
+      "speakers": 51872,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6747,11 +6747,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kabuverdianu",
-      "speakers": 530762,
-      "family": "Indo-European",
+      "language_name": "Northern Sami",
+      "speakers": 51530,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6759,8 +6759,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kenyang",
-      "speakers": 69362,
+      "language_name": "Anii",
+      "speakers": 51507,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -6771,9 +6771,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Koro",
-      "speakers": 63207,
-      "family": "Mande",
+      "language_name": "Kaingang",
+      "speakers": 50812,
+      "family": "Nuclear-Macro-Je",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6783,8 +6783,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kachhi",
-      "speakers": 994568,
+      "language_name": "Kashubian",
+      "speakers": 49767,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -6795,8 +6795,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Mankanya",
-      "speakers": 83151,
+      "language_name": "Koro Wachi",
+      "speakers": 46718,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -6807,9 +6807,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Komi-Permyak",
-      "speakers": 63775,
-      "family": "Uralic",
+      "language_name": "Chamorro",
+      "speakers": 46325,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6817,11 +6817,11 @@
       "language_modeling_chrf": 0.0,
       "translation_bleu": 0.0,
       "translation_chrf": 0.0
-    },
-    {
-      "language_name": "Jumli",
-      "speakers": 970493,
-      "family": "Indo-European",
+    },
+    {
+      "language_name": "Atsam",
+      "speakers": 44946,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6831,9 +6831,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Wadiyara Koli",
-      "speakers": 256851,
-      "family": "Indo-European",
+      "language_name": "Bushi",
+      "speakers": 44620,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6843,11 +6843,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kyrgyz",
-      "speakers": 3338267,
-      "family": "Turkic",
+      "language_name": "Romansh",
+      "speakers": 42020,
+      "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6855,9 +6855,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Latin",
-      "speakers": 820,
-      "family": "Indo-European",
+      "language_name": "Aghem",
+      "speakers": 38843,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6867,9 +6867,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ladino",
-      "speakers": 112781,
-      "family": "Indo-European",
+      "language_name": "Kuvi",
+      "speakers": 38457,
+      "family": "Dravidian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6879,8 +6879,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Langi",
-      "speakers": 509409,
+      "language_name": "Noon",
+      "speakers": 37767,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -6891,11 +6891,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Panjabi",
-      "speakers": 93433552,
+      "language_name": "Friulian",
+      "speakers": 37442,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6903,9 +6903,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lango (Uganda)",
-      "speakers": 1643614,
-      "family": "Nilotic",
+      "language_name": "Thulung",
+      "speakers": 36393,
+      "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6915,11 +6915,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Luxembourgish",
-      "speakers": 421015,
+      "language_name": "Emilian",
+      "speakers": 31201,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6927,9 +6927,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lak",
-      "speakers": 110543,
-      "family": "Nakh-Daghestanian",
+      "language_name": "Chuukese",
+      "speakers": 30731,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6939,9 +6939,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Tolaki",
-      "speakers": 347134,
-      "family": "Austronesian",
+      "language_name": "Western Mari",
+      "speakers": 29762,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6951,9 +6951,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Lawa",
-      "speakers": 87751,
-      "family": "Austroasiatic",
+      "language_name": "Hawaiian",
+      "speakers": 29605,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6963,8 +6963,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lepcha",
-      "speakers": 79743,
+      "language_name": "Mru",
+      "speakers": 29277,
       "family": "Sino-Tibetan",
       "average": 0.0,
       "in_benchmark": false,
@@ -6975,9 +6975,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lezghian",
-      "speakers": 255100,
-      "family": "Nakh-Daghestanian",
+      "language_name": "Cajun French",
+      "speakers": 27942,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -6987,11 +6987,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ganda",
-      "speakers": 5622890,
-      "family": "Atlantic-Congo",
+      "language_name": "Nheengatu",
+      "speakers": 26171,
+      "family": "Tupian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -6999,11 +6999,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Limburgish",
-      "speakers": 950422,
+      "language_name": "Aragonese",
+      "speakers": 26008,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7011,9 +7011,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Limbu",
-      "speakers": 368085,
-      "family": "Sino-Tibetan",
+      "language_name": "Cherokee",
+      "speakers": 25613,
+      "family": "Iroquoian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7023,11 +7023,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ligurian",
-      "speakers": 536663,
+      "language_name": "Sinte Romani",
+      "speakers": 24372,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7035,9 +7035,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Abkhazian",
-      "speakers": 91953,
-      "family": "Abkhaz-Adyge",
+      "language_name": "Ojibwa",
+      "speakers": 23747,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7047,9 +7047,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kuvi",
-      "speakers": 38457,
-      "family": "Dravidian",
+      "language_name": "Pohnpeian",
+      "speakers": 23560,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7059,9 +7059,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern Khmer",
-      "speakers": 1172616,
-      "family": "Austroasiatic",
+      "language_name": "Laz",
+      "speakers": 22965,
+      "family": "Kartvelian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7071,8 +7071,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Konkani",
-      "speakers": 4906533,
+      "language_name": "Muslim Tat",
+      "speakers": 22453,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -7083,9 +7083,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kwakʼwala",
-      "speakers": 377,
-      "family": "Wakashan",
+      "language_name": "Central Yupik",
+      "speakers": 20956,
+      "family": "Eskimo-Aleut",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7095,9 +7095,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kosraean",
-      "speakers": 7990,
-      "family": "Austronesian",
+      "language_name": "Dakota",
+      "speakers": 20832,
+      "family": "Siouan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7107,9 +7107,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kpelle",
-      "speakers": 1186303,
-      "family": "Mande",
+      "language_name": "Southern Altai",
+      "speakers": 19841,
+      "family": "Turkic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7119,9 +7119,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Karachay-Balkar",
-      "speakers": 240927,
-      "family": "Turkic",
+      "language_name": "Amo",
+      "speakers": 18620,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7131,9 +7131,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Krio",
-      "speakers": 6293684,
-      "family": "Indo-European",
+      "language_name": "Guajajára",
+      "speakers": 17784,
+      "family": "Tupian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7143,8 +7143,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kinaray-a",
-      "speakers": 425806,
+      "language_name": "Hanunoo",
+      "speakers": 17469,
       "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
@@ -7155,9 +7155,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Karelian",
-      "speakers": 116212,
-      "family": "Uralic",
+      "language_name": "Tsakhur",
+      "speakers": 16329,
+      "family": "Nakh-Daghestanian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7167,9 +7167,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kurukh",
-      "speakers": 2519571,
-      "family": "Dravidian",
+      "language_name": "Palauan",
+      "speakers": 16047,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7179,8 +7179,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kashmiri",
-      "speakers": 5598085,
+      "language_name": "Sanskrit",
+      "speakers": 15913,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": true,
@@ -7191,8 +7191,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Shambala",
-      "speakers": 995398,
+      "language_name": "Bassari",
+      "speakers": 15264,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -7203,9 +7203,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Bafia",
-      "speakers": 88784,
-      "family": "Atlantic-Congo",
+      "language_name": "Oji-Cree",
+      "speakers": 15078,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7215,9 +7215,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Colognian",
-      "speakers": 240479,
-      "family": "Indo-European",
+      "language_name": "Khamti",
+      "speakers": 13527,
+      "family": "Tai-Kadai",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7227,11 +7227,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kurdish",
-      "speakers": 6866757,
+      "language_name": "Upper Sorbian",
+      "speakers": 12826,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7239,9 +7239,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kumyk",
-      "speakers": 283444,
-      "family": "Turkic",
+      "language_name": "Chipewyan",
+      "speakers": 12816,
+      "family": "Athabaskan-Eyak-Tlingit",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7251,9 +7251,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Komi",
-      "speakers": 255100,
-      "family": "Uralic",
+      "language_name": "Innu-aimun",
+      "speakers": 12062,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7263,9 +7263,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kerinci",
-      "speakers": 373836,
-      "family": "Austronesian",
+      "language_name": "Lower Silesian",
+      "speakers": 11868,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7275,8 +7275,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Parkari Koli",
-      "speakers": 373602,
+      "language_name": "Walser",
+      "speakers": 11377,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -7287,9 +7287,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Cornish",
-      "speakers": 1973,
-      "family": "Indo-European",
+      "language_name": "Choctaw",
+      "speakers": 10977,
+      "family": "Muskogean",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7299,9 +7299,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Jutish",
-      "speakers": 0,
-      "family": "Indo-European",
+      "language_name": "Tagbanwa",
+      "speakers": 10045,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7311,9 +7311,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Machame",
-      "speakers": 433291,
-      "family": "Atlantic-Congo",
+      "language_name": "Xavánte",
+      "speakers": 9951,
+      "family": "Nuclear-Macro-Je",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7323,9 +7323,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Arpitan",
-      "speakers": 63777,
-      "family": "Indo-European",
+      "language_name": "Tuvalu",
+      "speakers": 9868,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7335,8 +7335,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Kachi Koli",
-      "speakers": 256851,
+      "language_name": "Northern Frisian",
+      "speakers": 9619,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -7345,13 +7345,13 @@
       "language_modeling_chrf": 0.0,
       "translation_bleu": 0.0,
       "translation_chrf": 0.0
-    },
-    {
-      "language_name": "Galician",
-      "speakers": 3515530,
-      "family": "Indo-European",
+    },
+    {
+      "language_name": "Roviana",
+      "speakers": 9591,
+      "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7359,9 +7359,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gilaki",
-      "speakers": 3906472,
-      "family": "Indo-European",
+      "language_name": "Wallisian",
+      "speakers": 9512,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7371,11 +7371,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Guarani",
-      "speakers": 5827107,
-      "family": "Tupian",
+      "language_name": "Tavringer Romani",
+      "speakers": 9488,
+      "family": "Speech Register",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7383,11 +7383,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Goan Konkani",
-      "speakers": 4243488,
-      "family": "Indo-European",
+      "language_name": "Cree",
+      "speakers": 9047,
+      "family": "Algic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7395,9 +7395,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gondi",
-      "speakers": 3182616,
-      "family": "Dravidian",
+      "language_name": "Kwasio",
+      "speakers": 8878,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7407,9 +7407,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gorontalo",
-      "speakers": 1094807,
-      "family": "Austronesian",
+      "language_name": "Lakota",
+      "speakers": 8316,
+      "family": "Siouan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7419,9 +7419,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gronings",
-      "speakers": 622094,
-      "family": "Indo-European",
+      "language_name": "Kosraean",
+      "speakers": 7990,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7431,9 +7431,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Garo",
-      "speakers": 821563,
-      "family": "Sino-Tibetan",
+      "language_name": "Inupiaq",
+      "speakers": 7983,
+      "family": "Eskimo-Aleut",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7443,8 +7443,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Swiss German",
-      "speakers": 7956952,
+      "language_name": "Zoroastrian Dari",
+      "speakers": 7983,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -7455,11 +7455,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gujarati",
-      "speakers": 61721799,
-      "family": "Indo-European",
+      "language_name": "Tasawaq",
+      "speakers": 7970,
+      "family": "Songhay",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7467,9 +7467,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Guajajára",
-      "speakers": 17784,
-      "family": "Tupian",
+      "language_name": "Buhid",
+      "speakers": 7970,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7479,9 +7479,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Wayuu",
-      "speakers": 132529,
-      "family": "Arawakan",
+      "language_name": "Mi'kmaw",
+      "speakers": 7916,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7491,9 +7491,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Frafra",
-      "speakers": 1026907,
-      "family": "Atlantic-Congo",
+      "language_name": "Hassaniyya",
+      "speakers": 7239,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7503,9 +7503,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gusii",
-      "speakers": 2622867,
-      "family": "Atlantic-Congo",
+      "language_name": "Lower Sorbian",
+      "speakers": 6974,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7515,9 +7515,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Manx",
-      "speakers": 1719,
-      "family": "Indo-European",
+      "language_name": "Nauru",
+      "speakers": 6930,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7527,9 +7527,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gurung",
-      "speakers": 87951,
-      "family": "Sino-Tibetan",
+      "language_name": "Eastern Lawa",
+      "speakers": 6898,
+      "family": "Austroasiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7539,9 +7539,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gwichʼin",
-      "speakers": 302,
-      "family": "Athabaskan-Eyak-Tlingit",
+      "language_name": "Yapese",
+      "speakers": 6556,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7551,9 +7551,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gujari",
-      "speakers": 467002,
-      "family": "Indo-European",
+      "language_name": "Atikamekw",
+      "speakers": 6408,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7563,9 +7563,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gilbertese",
-      "speakers": 67078,
-      "family": "Austronesian",
+      "language_name": "Piedmontese",
+      "speakers": 6178,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7575,9 +7575,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hakka Chinese",
-      "speakers": 32062460,
-      "family": "Sino-Tibetan",
+      "language_name": "Kalo Finnish Romani",
+      "speakers": 5015,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7587,9 +7587,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Geez",
-      "speakers": 0,
-      "family": "Afro-Asiatic",
+      "language_name": "Siksiká",
+      "speakers": 4900,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7599,9 +7599,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern Frisian",
-      "speakers": 9619,
-      "family": "Indo-European",
+      "language_name": "East Futuna",
+      "speakers": 4756,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7611,9 +7611,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Eastern Frisian",
-      "speakers": 2004,
-      "family": "Indo-European",
+      "language_name": "Taroko",
+      "speakers": 4721,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7623,9 +7623,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "East Futuna",
-      "speakers": 4756,
-      "family": "Austronesian",
+      "language_name": "Plains Cree",
+      "speakers": 4146,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7635,8 +7635,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Central-Eastern Niger Fulfulde",
-      "speakers": 1594068,
+      "language_name": "Safaliba",
+      "speakers": 4108,
       "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
@@ -7647,11 +7647,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Friulian",
-      "speakers": 37442,
-      "family": "Indo-European",
+      "language_name": "Muscogee",
+      "speakers": 3992,
+      "family": "Muskogean",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7659,11 +7659,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Nigerian Fulfulde",
-      "speakers": 14339876,
-      "family": "Atlantic-Congo",
+      "language_name": "Veps",
+      "speakers": 3543,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7671,9 +7671,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Fur",
-      "speakers": 1230163,
-      "family": "Furan",
+      "language_name": "Ménik",
+      "speakers": 3305,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7683,8 +7683,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Western Frisian",
-      "speakers": 743057,
+      "language_name": "Virgin Islands Creole English",
+      "speakers": 3113,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -7695,11 +7695,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Irish",
-      "speakers": 1237487,
-      "family": "Indo-European",
+      "language_name": "Turoyo",
+      "speakers": 3035,
+      "family": "Afro-Asiatic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7707,9 +7707,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ga",
-      "speakers": 821526,
-      "family": "Atlantic-Congo",
+      "language_name": "Ulithian",
+      "speakers": 2971,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7719,9 +7719,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gagauz",
-      "speakers": 111028,
-      "family": "Turkic",
+      "language_name": "Rotuman",
+      "speakers": 2527,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7731,9 +7731,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gan Chinese",
-      "speakers": 23698340,
-      "family": "Sino-Tibetan",
+      "language_name": "Warlpiri",
+      "speakers": 2496,
+      "family": "Pama-Nyungan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7743,9 +7743,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Gayo",
-      "speakers": 320431,
-      "family": "Austronesian",
+      "language_name": "Yangben",
+      "speakers": 2303,
+      "family": "Atlantic-Congo",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7755,9 +7755,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Garhwali",
-      "speakers": 3580443,
-      "family": "Indo-European",
+      "language_name": "Slave",
+      "speakers": 2299,
+      "family": "Athabaskan-Eyak-Tlingit",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7767,9 +7767,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Zoroastrian Dari",
-      "speakers": 7983,
-      "family": "Indo-European",
+      "language_name": "Dogrib",
+      "speakers": 2111,
+      "family": "Athabaskan-Eyak-Tlingit",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7779,8 +7779,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Guianese Creole French",
-      "speakers": 51872,
+      "language_name": "Eastern Frisian",
+      "speakers": 2004,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -7791,11 +7791,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Scottish Gaelic",
-      "speakers": 72337,
+      "language_name": "Cornish",
+      "speakers": 1973,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7803,11 +7803,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hausa",
-      "speakers": 40411882,
-      "family": "Afro-Asiatic",
+      "language_name": "Swampy Cree",
+      "speakers": 1809,
+      "family": "Algic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7815,9 +7815,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hawaiian",
-      "speakers": 29605,
-      "family": "Austronesian",
+      "language_name": "Mohawk",
+      "speakers": 1772,
+      "family": "Iroquoian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7827,9 +7827,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ngomba",
-      "speakers": 94333,
-      "family": "Atlantic-Congo",
+      "language_name": "Manx",
+      "speakers": 1719,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7839,9 +7839,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Interlingua",
-      "speakers": 136,
-      "family": "Artificial Language",
+      "language_name": "Lule Sami",
+      "speakers": 1530,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7851,9 +7851,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ibibio",
-      "speakers": 2996392,
-      "family": "Atlantic-Congo",
+      "language_name": "Naskapi",
+      "speakers": 1395,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7863,11 +7863,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Indonesian",
-      "speakers": 171207687,
+      "language_name": "Tokelau",
+      "speakers": 1285,
       "family": "Austronesian",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7875,9 +7875,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Interlingue",
-      "speakers": 1,
-      "family": "Artificial Language",
+      "language_name": "Niuean",
+      "speakers": 1120,
+      "family": "Austronesian",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7887,9 +7887,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ifè",
-      "speakers": 111910,
-      "family": "Atlantic-Congo",
+      "language_name": "Saterland Frisian",
+      "speakers": 962,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7899,11 +7899,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Igbo",
-      "speakers": 27823640,
-      "family": "Atlantic-Congo",
+      "language_name": "Seri",
+      "speakers": 901,
+      "family": null,
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7911,9 +7911,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Sichuan Yi",
-      "speakers": 8364120,
-      "family": "Sino-Tibetan",
+      "language_name": "Chilcotin",
+      "speakers": 867,
+      "family": "Athabaskan-Eyak-Tlingit",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7923,9 +7923,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Inupiaq",
-      "speakers": 7983,
-      "family": "Eskimo-Aleut",
+      "language_name": "Latin",
+      "speakers": 820,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7935,11 +7935,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Iloko",
-      "speakers": 10481376,
-      "family": "Austronesian",
+      "language_name": "Halkomelem",
+      "speakers": 716,
+      "family": "Salishan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7947,9 +7947,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ingush",
-      "speakers": 226755,
-      "family": "Nakh-Daghestanian",
+      "language_name": "Michif",
+      "speakers": 678,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7959,9 +7959,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ido",
-      "speakers": 0,
-      "family": "Artificial Language",
+      "language_name": "Skolt Sami",
+      "speakers": 613,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -7971,11 +7971,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Icelandic",
-      "speakers": 350734,
-      "family": "Indo-European",
+      "language_name": "Inari Sami",
+      "speakers": 613,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7983,11 +7983,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Italian",
-      "speakers": 70247060,
-      "family": "Indo-European",
+      "language_name": "Lillooet",
+      "speakers": 528,
+      "family": "Salishan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -7995,9 +7995,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Inuktitut",
-      "speakers": 90466,
-      "family": "Eskimo-Aleut",
+      "language_name": "Okanagan",
+      "speakers": 490,
+      "family": "Salishan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8007,9 +8007,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ingrian",
-      "speakers": 142,
-      "family": "Uralic",
+      "language_name": "Maliseet-Passamaquoddy",
+      "speakers": 490,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8019,11 +8019,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Japanese",
-      "speakers": 119729026,
-      "family": "Japonic",
+      "language_name": "Kwakʼwala",
+      "speakers": 377,
+      "family": "Wakashan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8031,9 +8031,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Jamaican Creole English",
-      "speakers": 2668142,
-      "family": "Indo-European",
+      "language_name": "Northern East Cree",
+      "speakers": 377,
+      "family": "Algic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8043,9 +8043,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Lojban",
-      "speakers": 0,
-      "family": "Artificial Language",
+      "language_name": "Gwichʼin",
+      "speakers": 302,
+      "family": "Athabaskan-Eyak-Tlingit",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8055,11 +8055,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Iban",
-      "speakers": 816302,
-      "family": "Bookkeeping",
+      "language_name": "Esperanto",
+      "speakers": 301,
+      "family": "Artificial Language",
       "average": 0.0,
-      "in_benchmark": false,
+      "in_benchmark": true,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8067,9 +8067,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Herero",
-      "speakers": 239336,
-      "family": "Atlantic-Congo",
+      "language_name": "Southern Sami",
+      "speakers": 296,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8079,8 +8079,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hazaragi",
-      "speakers": 2161984,
+      "language_name": "Tsakonian",
+      "speakers": 202,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -8091,11 +8091,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Armenian",
-      "speakers": 5317273,
-      "family": "Indo-European",
+      "language_name": "Ingrian",
+      "speakers": 142,
+      "family": "Uralic",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8103,11 +8103,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hebrew",
-      "speakers": 8675480,
-      "family": "Afro-Asiatic",
+      "language_name": "Interlingua",
+      "speakers": 136,
+      "family": "Artificial Language",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8115,9 +8115,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Fiji Hindi",
-      "speakers": 383749,
-      "family": "Indo-European",
+      "language_name": "Araona",
+      "speakers": 105,
+      "family": "Pano-Tacanan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8127,9 +8127,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hiligaynon",
-      "speakers": 9171204,
-      "family": "Austronesian",
+      "language_name": "Prussian",
+      "speakers": 38,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8139,9 +8139,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Southern Hindko",
-      "speakers": 957354,
-      "family": "Indo-European",
+      "language_name": "Interlingue",
+      "speakers": 1,
+      "family": "Artificial Language",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8151,11 +8151,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Chhattisgarhi",
-      "speakers": 14586990,
-      "family": "Indo-European",
+      "language_name": "Literary Chinese",
+      "speakers": 0,
+      "family": "Sino-Tibetan",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8163,9 +8163,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hmong Njua",
-      "speakers": 781687,
-      "family": "Hmong-Mien",
+      "language_name": "Ido",
+      "speakers": 0,
+      "family": "Artificial Language",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8175,9 +8175,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hanunoo",
-      "speakers": 17469,
-      "family": "Austronesian",
+      "language_name": "Lojban",
+      "speakers": 0,
+      "family": "Artificial Language",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8187,8 +8187,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Northern Hindko",
-      "speakers": 3969517,
+      "language_name": "Jutish",
+      "speakers": 0,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -8199,9 +8199,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hiri Motu",
-      "speakers": 152449,
-      "family": "Pidgin",
+      "language_name": "Votic",
+      "speakers": 0,
+      "family": "Uralic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8211,9 +8211,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Ho",
-      "speakers": 1312829,
-      "family": "Austroasiatic",
+      "language_name": "Geez",
+      "speakers": 0,
+      "family": "Afro-Asiatic",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8223,9 +8223,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hadothi",
-      "speakers": 1087394,
-      "family": "Indo-European",
+      "language_name": "Osage",
+      "speakers": 0,
+      "family": "Siouan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8235,11 +8235,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Croatian",
-      "speakers": 6813164,
+      "language_name": "Romagnol",
+      "speakers": 0,
       "family": "Indo-European",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8247,8 +8247,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Upper Sorbian",
-      "speakers": 12826,
+      "language_name": "Church Slavic",
+      "speakers": 0,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -8259,9 +8259,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Xiang Chinese",
-      "speakers": 40426580,
-      "family": "Sino-Tibetan",
+      "language_name": "Samogitian",
+      "speakers": 0,
+      "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8271,11 +8271,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Haitian Creole",
-      "speakers": 8964918,
-      "family": "Indo-European",
+      "language_name": "Obolo",
+      "speakers": 0,
+      "family": "Atlantic-Congo",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8283,11 +8283,11 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Hungarian",
-      "speakers": 12443430,
-      "family": "Uralic",
+      "language_name": "Chickasaw",
+      "speakers": 0,
+      "family": "Muskogean",
       "average": 0.0,
-      "in_benchmark": true,
+      "in_benchmark": false,
       "NaN": 0.0,
       "classification_accuracy": 0.0,
       "language_modeling_chrf": 0.0,
@@ -8295,9 +8295,9 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Halkomelem",
-      "speakers": 716,
-      "family": "Salishan",
+      "language_name": "Caddo",
+      "speakers": 0,
+      "family": "Caddoan",
       "average": 0.0,
       "in_benchmark": false,
       "NaN": 0.0,
@@ -8307,8 +8307,8 @@
       "translation_chrf": 0.0
     },
     {
-      "language_name": "Zaza",
-      "speakers": 1148245,
+      "language_name": "Palatine German",
+      "speakers": 0,
       "family": "Indo-European",
       "average": 0.0,
       "in_benchmark": false,
@@ -8323,8 +8323,9 @@
     {
       "name": "FLORES+",
       "author": "Meta",
+      "author_url": "https://ai.meta.com",
       "url": "https://huggingface.co/datasets/openlanguagedata/flores_plus",
-      "n_languages": 200.0,
+      "n_languages": 200,
       "tasks": [
         "translation",
         "classification",
@@ -8337,8 +8338,9 @@
     {
       "name": "FLEURS",
       "author": "Meta",
+      "author_url": "https://ai.meta.com",
       "url": "https://huggingface.co/datasets/google/fleurs",
-      "n_languages": 102.0,
+      "n_languages": 102,
       "tasks": [
         "speech_recognition"
       ],
@@ -8349,8 +8351,9 @@
     {
       "name": "CommonVoice",
       "author": "Mozilla",
+      "author_url": "https://mozilla.ai",
       "url": "https://huggingface.co/datasets/mozilla-foundation/common_voice_17_0",
-      "n_languages": 124.0,
+      "n_languages": 124,
       "tasks": [
         "speech_recognition"
       ],
@@ -8361,8 +8364,9 @@
     {
       "name": "MMMLU",
       "author": "OpenAI",
+      "author_url": "https://openai.com",
       "url": "https://huggingface.co/datasets/openai/MMMLU",
-      "n_languages": 14.0,
+      "n_languages": 14,
       "tasks": [
         "question_answering"
       ],
@@ -8373,8 +8377,9 @@
     {
       "name": "AfriMMLU",
       "author": "Masakhane",
+      "author_url": "https://www.masakhane.io",
       "url": "https://huggingface.co/datasets/masakhane/afrimmlu",
-      "n_languages": 17.0,
+      "n_languages": 17,
       "tasks": [
         "question_answering"
       ],
@@ -8384,9 +8389,10 @@
     },
     {
       "name": "Okapi MMLU",
-      "author": "Okapi",
+      "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/jon-tow/okapi_mmlu",
-      "n_languages": 16.0,
+      "n_languages": 16,
       "tasks": [
         "question_answering"
       ],
@@ -8397,8 +8403,9 @@
     {
       "name": "Global MMLU",
       "author": "Cohere",
+      "author_url": "https://cohere.com",
       "url": "https://huggingface.co/datasets/CohereForAI/Global-MMLU",
-      "n_languages": 42.0,
+      "n_languages": 42,
       "tasks": [
         "question_answering"
       ],
@@ -8409,8 +8416,9 @@
     {
       "name": "MGSM",
       "author": "Google",
+      "author_url": "https://google.com",
       "url": "https://huggingface.co/datasets/juletxara/mgsm",
-      "n_languages": 10.0,
+      "n_languages": 10,
       "tasks": [
         "math"
       ],
@@ -8421,8 +8429,9 @@
     {
       "name": "AfriMGSM",
       "author": "Masakhane",
+      "author_url": "https://www.masakhane.io",
       "url": "https://huggingface.co/datasets/masakhane/afrimgsm",
-      "n_languages": 18.0,
+      "n_languages": 18,
       "tasks": [
         "math"
       ],
@@ -8432,9 +8441,10 @@
     },
     {
       "name": "Okapi ARC Challenge",
-      "author": "Okapi",
+      "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/jon-tow/okapi_arc_challenge",
-      "n_languages": 31.0,
+      "n_languages": 31,
       "tasks": [
         "question_answering"
       ],
@@ -8445,8 +8455,9 @@
     {
       "name": "Uhuru ARC Easy",
       "author": "Masakhane",
+      "author_url": "https://www.masakhane.io",
       "url": "https://huggingface.co/datasets/masakhane/uhura-arc-easy",
-      "n_languages": 6.0,
+      "n_languages": 6,
       "tasks": [
         "question_answering"
       ],
@@ -8456,9 +8467,10 @@
     },
     {
       "name": "Okapi TruthfulQA",
-      "author": "Okapi",
+      "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/jon-tow/okapi_truthfulqa/tree/main/data",
-      "n_languages": 31.0,
+      "n_languages": 31,
       "tasks": [
         "question_answering"
       ],
@@ -8469,8 +8481,9 @@
     {
       "name": "Uhura TruthfulQA",
       "author": "Masakhane",
+      "author_url": "https://www.masakhane.io",
       "url": "https://huggingface.co/datasets/masakhane/uhura-truthfulqa",
-      "n_languages": 6.0,
+      "n_languages": 6,
       "tasks": [
         "question_answering"
       ],
@@ -8481,8 +8494,9 @@
     {
       "name": "XNLI",
       "author": "Meta",
+      "author_url": "https://ai.meta.com",
       "url": "https://huggingface.co/datasets/facebook/xnli",
-      "n_languages": 14.0,
+      "n_languages": 14,
       "tasks": [
         "classification"
       ],
@@ -8493,8 +8507,9 @@
     {
       "name": "AfriXNLI",
       "author": "Masakhane",
+      "author_url": "https://www.masakhane.io",
       "url": "https://huggingface.co/datasets/masakhane/afrixnli",
-      "n_languages": 18.0,
+      "n_languages": 18,
       "tasks": [
         "classification"
       ],
@@ -8504,9 +8519,10 @@
     },
     {
       "name": "Okapi HellaSwag",
-      "author": "Okapi",
+      "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/jon-tow/okapi_hellaswag",
-      "n_languages": 31.0,
+      "n_languages": 31,
       "tasks": [
         "question_answering"
       ],
@@ -8517,8 +8533,9 @@
     {
       "name": "WikiANN / PAN-X",
       "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/unimelb-nlp/wikiann",
-      "n_languages": 176.0,
+      "n_languages": 176,
       "tasks": [
         "ner"
       ],
@@ -8529,8 +8546,9 @@
     {
       "name": "MSVAMP",
       "author": "Microsoft",
+      "author_url": "https://microsoft.com",
       "url": "https://huggingface.co/datasets/Mathoctopus/MSVAMP",
-      "n_languages": 10.0,
+      "n_languages": 10,
       "tasks": [
         "math"
       ],
@@ -8541,8 +8559,9 @@
     {
       "name": "XLSUM",
       "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/csebuetnlp/xlsum",
-      "n_languages": 45.0,
+      "n_languages": 45,
       "tasks": [
         "summarization"
       ],
@@ -8553,8 +8572,9 @@
     {
       "name": "SEA-IFEVAL",
       "author": "AI Singapore",
+      "author_url": "https://aisingapore.org",
       "url": "https://huggingface.co/datasets/aisingapore/instruction_following-ifeval",
-      "n_languages": 7.0,
+      "n_languages": 7,
       "tasks": [
         "instruction_following"
       ],
@@ -8565,8 +8585,9 @@
     {
       "name": "XTREME",
       "author": "Google",
+      "author_url": "https://google.com",
       "url": "https://huggingface.co/datasets/google/xtreme",
-      "n_languages": 40.0,
+      "n_languages": 40,
       "tasks": [
         "translation",
         "classification",
@@ -8580,8 +8601,9 @@
     {
       "name": "XGLUE",
       "author": "Microsoft",
+      "author_url": "https://microsoft.com",
       "url": "https://huggingface.co/datasets/microsoft/xglue",
-      "n_languages": 18.0,
+      "n_languages": 18,
       "tasks": [
         "pos"
       ],
@@ -8592,8 +8614,9 @@
     {
       "name": "IndicGLUE",
       "author": "AI4Bharat",
+      "author_url": "https://models.ai4bharat.org",
       "url": "https://huggingface.co/datasets/ai4bharat/indic_glue",
-      "n_languages": 11.0,
+      "n_languages": 11,
       "tasks": [
         "question_answering"
       ],
@@ -8604,8 +8627,9 @@
     {
       "name": "Opus Gnome",
       "author": "Helsinki NLP",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/Helsinki-NLP/opus_gnome",
-      "n_languages": 187.0,
+      "n_languages": 187,
       "tasks": [
         "translation"
       ],
@@ -8616,8 +8640,9 @@
     {
       "name": "Opus Paracrawl",
       "author": "Helsinki NLP",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/Helsinki-NLP/opus_paracrawl",
-      "n_languages": 43.0,
+      "n_languages": 43,
       "tasks": [
         "translation"
       ],
@@ -8628,8 +8653,9 @@
     {
       "name": "CCAligned",
       "author": "Meta",
+      "author_url": "https://ai.meta.com",
       "url": "https://huggingface.co/datasets/ahelk/ccaligned_multilingual",
-      "n_languages": 137.0,
+      "n_languages": 137,
       "tasks": [
         "translation"
       ],
@@ -8640,8 +8666,9 @@
     {
       "name": "OPUS Collection",
       "author": "Helsinki NLP",
-      "url": "https://opus.nlpl.eu/",
-      "n_languages": 747.0,
+      "author_url": null,
+      "url": "https://opus.nlpl.eu",
+      "n_languages": 747,
       "tasks": [
         "translation"
       ],
@@ -8652,8 +8679,9 @@
     {
       "name": "MasakhaNER",
       "author": "Masakhane",
+      "author_url": "https://www.masakhane.io",
       "url": "https://huggingface.co/datasets/masakhane/masakhaner",
-      "n_languages": 10.0,
+      "n_languages": 10,
       "tasks": [
         "ner"
       ],
@@ -8663,9 +8691,10 @@
     },
     {
       "name": "Multilingual Sentiments",
-      "author": null,
+      "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/tyqiangz/multilingual-sentiments",
-      "n_languages": 12.0,
+      "n_languages": 12,
       "tasks": [
         "sentiment_analysis"
       ],
@@ -8676,8 +8705,9 @@
     {
       "name": "CulturaX",
       "author": "Academic",
+      "author_url": null,
       "url": "https://huggingface.co/datasets/uonlp/CulturaX",
-      "n_languages": 167.0,
+      "n_languages": 167,
       "tasks": [
         "language_modeling"
       ],
@@ -8688,8 +8718,9 @@
     {
       "name": "Tülu 3 SFT Mixture",
       "author": "AllenAI",
+      "author_url": "https://allenai.org",
       "url": "https://huggingface.co/datasets/allenai/tulu-3-sft-mixture",
-      "n_languages": 70.0,
+      "n_languages": 70,
       "tasks": [
         "instruction_following"
       ],
@@ -8700,8 +8731,9 @@
     {
       "name": "xP3",
       "author": "BigScience",
+      "author_url": "https://bigscience.huggingface.co",
       "url": "https://huggingface.co/datasets/bigscience/xP3",
-      "n_languages": 46.0,
+      "n_languages": 46,
       "tasks": [
         "instruction_following"
       ],
@@ -8712,8 +8744,9 @@
     {
       "name": "Aya",
       "author": "Cohere",
+      "author_url": "https://cohere.com",
       "url": "https://huggingface.co/datasets/CohereForAI/aya_dataset",
-      "n_languages": 65.0,
+      "n_languages": 65,
       "tasks": [
         "instruction_following"
       ],
@@ -8724,8 +8757,9 @@
     {
       "name": "Lanfrica",
       "author": "Lanfrica",
+      "author_url": "https://lanfrica.com",
       "url": "https://lanfrica.com/records?language=yor&task=machine%20translation",
-      "n_languages": 2200.0,
+      "n_languages": 2200,
       "tasks": [
         "datasets"
       ],
@@ -8736,8 +8770,9 @@
     {
       "name": "HuggingFace Languages",
       "author": "HuggingFace",
+      "author_url": "https://huggingface.co",
       "url": "https://huggingface.co/languages",
-      "n_languages": 4680.0,
+      "n_languages": 4680,
       "tasks": [
         "datasets",
         "models"
@@ -8749,8 +8784,9 @@
     {
       "name": "HuggingFace Multilingual Datasets",
       "author": "HuggingFace",
+      "author_url": "https://huggingface.co",
       "url": "https://huggingface.co/datasets?other=multilinguality:multilingual",
-      "n_languages": null,
+      "n_languages": 2012,
       "tasks": [
         "datasets"
       ],