Yasinjan99
/

wav2vec2-large-xls-r-300m-cv-11-ug-test

Automatic Speech Recognition

Inference Endpoints

Model card Files Files and versions Community

Yasinjan99 commited on Nov 10, 2023

Commit

eb335b3

•

1 Parent(s): a2139a7

Upload tokenizer

Files changed (3) hide show

added_tokens.json +2 -2
tokenizer_config.json +4 -4
vocab.json +44 -48

added_tokens.json CHANGED Viewed

@@ -1,4 +1,4 @@
 {
-  "</s>": 50,
-  "<s>": 49
 }

 {
+  "</s>": 46,
+  "<s>": 45
 }

tokenizer_config.json CHANGED Viewed

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "47": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": false
     },
-    "48": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": false
     },
-    "49": {
       "content": "<s>",
       "lstrip": false,
       "normalized": true,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "50": {
       "content": "</s>",
       "lstrip": false,
       "normalized": true,

 {
   "added_tokens_decoder": {
+    "43": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
       "single_word": false,
       "special": false
     },
+    "44": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
       "single_word": false,
       "special": false
     },
+    "45": {
       "content": "<s>",
       "lstrip": false,
       "normalized": true,
       "single_word": false,
       "special": true
     },
+    "46": {
       "content": "</s>",
       "lstrip": false,
       "normalized": true,

vocab.json CHANGED Viewed

@@ -1,51 +1,47 @@
 {
-  "[PAD]": 48,
-  "[UNK]": 47,
-  "`": 1,
   "|": 0,
-  "«": 2,
-  "»": 3,
-  "،": 4,
-  "؛": 5,
-  "؟": 6,
-  "ئ": 7,
-  "ا": 8,
-  "ب": 9,
-  "ت": 10,
-  "ج": 11,
-  "خ": 12,
-  "د": 13,
-  "ر": 14,
-  "ز": 15,
-  "س": 16,
-  "ش": 17,
-  "غ": 18,
-  "ف": 19,
-  "ق": 20,
-  "ك": 21,
-  "ل": 22,
-  "م": 23,
-  "ن": 24,
-  "و": 25,
-  "ى": 26,
-  "ي": 27,
-  "پ": 28,
-  "چ": 29,
-  "ژ": 30,
-  "ک": 31,
-  "ڭ": 32,
-  "گ": 33,
-  "ھ": 34,
-  "ۆ": 35,
-  "ۇ": 36,
-  "ۈ": 37,
-  "ۋ": 38,
-  "ی": 39,
-  "ې": 40,
-  "ە": 41,
-  "–": 42,
-  "—": 43,
-  "‹": 44,
-  "›": 45,
-  "−": 46
 }

 {
+  "[PAD]": 44,
+  "[UNK]": 43,
   "|": 0,
+  "«": 1,
+  "»": 2,
+  "،": 3,
+  "؛": 4,
+  "؟": 5,
+  "ئ": 6,
+  "ا": 7,
+  "ب": 8,
+  "ت": 9,
+  "ج": 10,
+  "خ": 11,
+  "د": 12,
+  "ر": 13,
+  "ز": 14,
+  "س": 15,
+  "ش": 16,
+  "غ": 17,
+  "ف": 18,
+  "ق": 19,
+  "ك": 20,
+  "ل": 21,
+  "م": 22,
+  "ن": 23,
+  "و": 24,
+  "ى": 25,
+  "ي": 26,
+  "پ": 27,
+  "چ": 28,
+  "ژ": 29,
+  "ڭ": 30,
+  "گ": 31,
+  "ھ": 32,
+  "ۆ": 33,
+  "ۇ": 34,
+  "ۈ": 35,
+  "ۋ": 36,
+  "ې": 37,
+  "ە": 38,
+  "–": 39,
+  "—": 40,
+  "‹": 41,
+  "›": 42
 }