ndeclarke
/

wav2vec2-mms-1b-CV17.0-training_set_variations

Automatic Speech Recognition

Generated from Trainer

Inference Endpoints

Model card Files Files and versions Metrics Training metrics Community

ndeclarke commited on 8 days ago

Commit

e41a29f

•

1 Parent(s): b196010

Upload tokenizer

Files changed (2) hide show

tokenizer_config.json +1 -2
vocab.json +0 -113

tokenizer_config.json CHANGED Viewed

@@ -39,9 +39,8 @@
   "eos_token": "</s>",
   "model_max_length": 1000000000000000019884624838656,
   "pad_token": "[PAD]",
-  "processor_class": "Wav2Vec2Processor",
   "replace_word_delimiter_char": " ",
-  "target_lang": "tam-512",
   "tokenizer_class": "Wav2Vec2CTCTokenizer",
   "unk_token": "[UNK]",
   "word_delimiter_token": "|"

   "eos_token": "</s>",
   "model_max_length": 1000000000000000019884624838656,
   "pad_token": "[PAD]",
   "replace_word_delimiter_char": " ",
+  "target_lang": "tam-32",
   "tokenizer_class": "Wav2Vec2CTCTokenizer",
   "unk_token": "[UNK]",
   "word_delimiter_token": "|"

vocab.json CHANGED Viewed

@@ -1,118 +1,5 @@
 {
-  "tam-128": {
-    "&": 1,
-    "[PAD]": 53,
-    "[UNK]": 52,
-    "_": 2,
-    "|": 0,
-    "ஃ": 3,
-    "அ": 4,
-    "ஆ": 5,
-    "இ": 6,
-    "ஈ": 7,
-    "உ": 8,
-    "ஊ": 9,
-    "எ": 10,
-    "ஏ": 11,
-    "ஐ": 12,
-    "ஒ": 13,
-    "ஓ": 14,
-    "ஔ": 15,
-    "க": 16,
-    "ங": 17,
-    "ச": 18,
-    "ஜ": 19,
-    "ஞ": 20,
-    "ட": 21,
-    "ண": 22,
-    "த": 23,
-    "ந": 24,
-    "ன": 25,
-    "ப": 26,
-    "ம": 27,
-    "ய": 28,
-    "ர": 29,
-    "ற": 30,
-    "ல": 31,
-    "ள": 32,
-    "ழ": 33,
-    "வ": 34,
-    "ஷ": 35,
-    "ஸ": 36,
-    "ஹ": 37,
-    "ா": 38,
-    "ி": 39,
-    "ீ": 40,
-    "ு": 41,
-    "ூ": 42,
-    "ெ": 43,
-    "ே": 44,
-    "ை": 45,
-    "ொ": 46,
-    "ோ": 47,
-    "ௌ": 48,
-    "்": 49,
-    "ௗ": 50,
-    "ഥ": 51
-  },
   "tam-32": {
-    "&": 1,
-    "[PAD]": 54,
-    "[UNK]": 53,
-    "_": 2,
-    "|": 0,
-    "¾": 3,
-    "ஃ": 4,
-    "அ": 5,
-    "ஆ": 6,
-    "இ": 7,
-    "ஈ": 8,
-    "உ": 9,
-    "ஊ": 10,
-    "எ": 11,
-    "ஏ": 12,
-    "ஐ": 13,
-    "ஒ": 14,
-    "ஓ": 15,
-    "ஔ": 16,
-    "க": 17,
-    "ங": 18,
-    "ச": 19,
-    "ஜ": 20,
-    "ஞ": 21,
-    "ட": 22,
-    "ண": 23,
-    "த": 24,
-    "ந": 25,
-    "ன": 26,
-    "ப": 27,
-    "ம": 28,
-    "ய": 29,
-    "ர": 30,
-    "ற": 31,
-    "ல": 32,
-    "ள": 33,
-    "ழ": 34,
-    "வ": 35,
-    "ஷ": 36,
-    "ஸ": 37,
-    "ஹ": 38,
-    "ா": 39,
-    "ி": 40,
-    "ீ": 41,
-    "ு": 42,
-    "ூ": 43,
-    "ெ": 44,
-    "ே": 45,
-    "ை": 46,
-    "ொ": 47,
-    "ோ": 48,
-    "ௌ": 49,
-    "்": 50,
-    "ௗ": 51,
-    "ഥ": 52
-  },
-  "tam-512": {
     "&": 1,
     "[PAD]": 53,
     "[UNK]": 52,

 {
   "tam-32": {
     "&": 1,
     "[PAD]": 53,
     "[UNK]": 52,