TiberiuCristianLeon commited on
Commit
08aee1f
·
verified ·
1 Parent(s): 496c579

Update languagecodes.py

Browse files
Files changed (1) hide show
  1. languagecodes.py +55 -0
languagecodes.py CHANGED
@@ -204,3 +204,58 @@ nllb_language_codes: dict[str, str] = {
204
  "Standard Malay": "zsm_Latn",
205
  "Zulu": "zul_Latn",
206
  }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
204
  "Standard Malay": "zsm_Latn",
205
  "Zulu": "zul_Latn",
206
  }
207
+ mbart_large_languages = {
208
+ 'Arabic': 'ar_AR',
209
+ 'Czech': 'cs_CZ',
210
+ 'German': 'de_DE',
211
+ 'English': 'en_XX',
212
+ 'Spanish': 'es_XX',
213
+ 'Estonian': 'et_EE',
214
+ 'Finnish': 'fi_FI',
215
+ 'French': 'fr_XX',
216
+ 'Gujarati': 'gu_IN',
217
+ 'Hindi': 'hi_IN',
218
+ 'Italian': 'it_IT',
219
+ 'Japanese': 'ja_XX',
220
+ 'Kazakh': 'kk_KZ',
221
+ 'Korean': 'ko_KR',
222
+ 'Lithuanian': 'lt_LT',
223
+ 'Latvian': 'lv_LV',
224
+ 'Burmese': 'my_MM',
225
+ 'Nepali': 'ne_NP',
226
+ 'Dutch': 'nl_XX',
227
+ 'Romanian': 'ro_RO',
228
+ 'Russian': 'ru_RU',
229
+ 'Sinhala': 'si_LK',
230
+ 'Turkish': 'tr_TR',
231
+ 'Vietnamese': 'vi_VN',
232
+ 'Chinese': 'zh_CN',
233
+ 'Afrikaans': 'af_ZA',
234
+ 'Azerbaijani': 'az_AZ',
235
+ 'Bengali': 'bn_IN',
236
+ 'Persian': 'fa_IR',
237
+ 'Hebrew': 'he_IL',
238
+ 'Croatian': 'hr_HR',
239
+ 'Indonesian': 'id_ID',
240
+ 'Georgian': 'ka_GE',
241
+ 'Khmer': 'km_KH',
242
+ 'Macedonian': 'mk_MK',
243
+ 'Malayalam': 'ml_IN',
244
+ 'Mongolian': 'mn_MN',
245
+ 'Marathi': 'mr_IN',
246
+ 'Polish': 'pl_PL',
247
+ 'Pashto': 'ps_AF',
248
+ 'Portuguese': 'pt_XX',
249
+ 'Swedish': 'sv_SE',
250
+ 'Swahili': 'sw_KE',
251
+ 'Tamil': 'ta_IN',
252
+ 'Telugu': 'te_IN',
253
+ 'Thai': 'th_TH',
254
+ 'Tagalog': 'tl_XX',
255
+ 'Ukrainian': 'uk_UA',
256
+ 'Urdu': 'ur_PK',
257
+ 'Xhosa': 'xh_ZA',
258
+ 'Galician': 'gl_ES',
259
+ 'Slovene': 'sl_SI'
260
+ }
261
+