CoolFace
Apppublic

bigscience-data/document-sizes

sourceHugging Faceupdated 2y agoView on Hugging Face
2likes
variables.py100 linesDownload Raw Back to root
1MAPPING_LANG_CODE_TO_TEXT = {2    "ar": "Arabic",3    "ca": "Catalan",4    "code": "code",5    "en": "English",6    "es": "Spanish",7    "eu": "Basque",8    "fr": "French",9    "id": "Indonesian",10    "indic-as": "Assamese",11    "indic-bn": "Bengali",12    "indic-gu": "Gujarati",13    "indic-hi": "Hindi",14    "indic-kn": "Kannada",15    "indic-ml": "Malayalam",16    "indic-mr": "Marathi",17    "indic-ne": "Nepali",18    "indic-or": "Odia",19    "indic-pa": "Punjabi",20    "indic-ta": "Tamil",21    "indic-te": "Telugu",22    "indic-ur": "Urdu",23    "nigercongo-ak": "Akan",24    "nigercongo-bm": "Bambara",25    "nigercongo-fon": "Fon",26    "nigercongo-ig": "Igbo",27    "nigercongo-ki": "Kikuyu",28    "nigercongo-lg": "Luganda",29    "nigercongo-ln": "Lingala",30    "nigercongo-nso": "Northern Sotho",31    "nigercongo-ny": "Chi Chewa",32    "nigercongo-rn": "Kirundi",33    "nigercongo-rw": "Kinyarwanda",34    "nigercongo-sn": "Chi Shona",35    "nigercongo-st": "Sesotho",36    "nigercongo-sw": "Swahili",37    "nigercongo-tn": "Setswana",38    "nigercongo-ts": "Xitsonga",39    "nigercongo-tum": "Chi Tumbuka",40    "nigercongo-tw": "Twi",41    "nigercongo-wo": "Wolof",42    "nigercongo-xh": "Xhosa",43    "nigercongo-yo": "Yoruba",44    "nigercongo-zu": "Isi Zulu",45    "pt": "Portuguese",46    "vi": "Vietnamese",47    "zhs": "Simplified Chinese",48    "zht": "Traditional Chinese",49}50 51PLOT_SIZES_PER_LANG = {52    "indic-ta": {"width": "7.2", "num_ds": "13"},53    "en": {"width": "19.6", "num_ds": "44"},54    "es": {"width": "52.0", "num_ds": "125"},55    "indic-kn": {"width": "5.2", "num_ds": "8"},56    "zht": {"width": "2.8", "num_ds": "2"},57    "nigercongo-ki": {"width": "2.4", "num_ds": "1"},58    "indic-pa": {"width": "5.6", "num_ds": "9"},59    "vi": {"width": "10.0", "num_ds": "20"},60    "zhs": {"width": "8.8", "num_ds": "17"},61    "fr": {"width": "10.8", "num_ds": "22"},62    "eu": {"width": "7.6000000000000005", "num_ds": "14"},63    "indic-te": {"width": "6.800000000000001", "num_ds": "12"},64    "indic-hi": {"width": "10.0", "num_ds": "20"},65    "pt": {"width": "9.600000000000001", "num_ds": "19"},66    "indic-bn": {"width": "8.4", "num_ds": "16"},67    "indic-mr": {"width": "6.4", "num_ds": "11"},68    "indic-gu": {"width": "6.0", "num_ds": "10"},69    "ca": {"width": "10.0", "num_ds": "20"},70    "id": {"width": "12.4", "num_ds": "26"},71    "ar": {"width": "12.0", "num_ds": "25"},72    "indic-or": {"width": "5.6", "num_ds": "9"},73    "indic-ur": {"width": "7.2", "num_ds": "13"},74    "nigercongo-ig": {"width": "2.4", "num_ds": "1"},75    "indic-as": {"width": "4.4", "num_ds": "6"},76    "indic-ml": {"width": "6.800000000000001", "num_ds": "12"},77    "nigercongo-ny": {"width": "2.4", "num_ds": "1"},78    "nigercongo-tw": {"width": "2.4", "num_ds": "1"},79    "nigercongo-rn": {"width": "2.4", "num_ds": "1"},80    "nigercongo-st": {"width": "2.4", "num_ds": "1"},81    "nigercongo-yo": {"width": "2.4", "num_ds": "1"},82    "nigercongo-ak": {"width": "2.4", "num_ds": "1"},83    "nigercongo-lg": {"width": "2.4", "num_ds": "1"},84    "nigercongo-bm": {"width": "2.4", "num_ds": "1"},85    "nigercongo-wo": {"width": "2.4", "num_ds": "1"},86    "nigercongo-ln": {"width": "2.4", "num_ds": "1"},87    "nigercongo-nso": {"width": "2.4", "num_ds": "1"},88    "code": {"width": "2.8", "num_ds": "2"},89    "indic-ne": {"width": "2.4", "num_ds": "1"},90    "nigercongo-ts": {"width": "2.4", "num_ds": "1"},91    "nigercongo-zu": {"width": "2.4", "num_ds": "1"},92    "nigercongo-sn": {"width": "2.4", "num_ds": "1"},93    "nigercongo-sw": {"width": "2.4", "num_ds": "1"},94    "nigercongo-tum": {"width": "2.4", "num_ds": "1"},95    "nigercongo-tn": {"width": "2.4", "num_ds": "1"},96    "nigercongo-xh": {"width": "2.4", "num_ds": "1"},97    "nigercongo-rw": {"width": "2.4", "num_ds": "1"},98    "nigercongo-fon": {"width": "2.4", "num_ds": "1"},99}100