Upload tokenizer

tokenizer.json

@@ -980,7 +980,7 @@
     },
     {
       "id": 32100,
-      "content": "ɑ",
+      "content": "ʤ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -989,7 +989,7 @@
     },
     {
       "id": 32101,
-      "content": "æ",
+      "content": "ɑ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -998,7 +998,7 @@
     },
     {
       "id": 32102,
-      "content": "ɐ",
+      "content": "æ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1007,7 +1007,7 @@
     },
     {
       "id": 32103,
-      "content": "̃",
+      "content": "ɐ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1016,7 +1016,7 @@
     },
     {
       "id": 32104,
-      "content": "β",
+      "content": "̃",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1025,7 +1025,7 @@
     },
     {
       "id": 32105,
-      "content": "ɓ",
+      "content": "β",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1034,7 +1034,7 @@
     },
     {
       "id": 32106,
-      "content": "ʙ",
+      "content": "ɓ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1043,7 +1043,7 @@
     },
     {
       "id": 32107,
-      "content": "ɕ",
+      "content": "ʙ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1052,7 +1052,7 @@
     },
     {
       "id": 32108,
-      "content": "ð",
+      "content": "ɕ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1061,7 +1061,7 @@
     },
     {
       "id": 32109,
-      "content": "͡",
+      "content": "ð",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1070,7 +1070,7 @@
     },
     {
       "id": 32110,
-      "content": "ʒ",
+      "content": "͡",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1079,7 +1079,7 @@
     },
     {
       "id": 32111,
-      "content": "ɖ",
+      "content": "ʒ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1088,7 +1088,7 @@
     },
     {
       "id": 32112,
-      "content": "ɗ",
+      "content": "ɖ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1097,7 +1097,7 @@
     },
     {
       "id": 32113,
-      "content": "ə",
+      "content": "ɗ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1106,7 +1106,7 @@
     },
     {
       "id": 32114,
-      "content": "ɚ",
+      "content": "ə",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1115,7 +1115,7 @@
     },
     {
       "id": 32115,
-      "content": "ɵ",
+      "content": "ɚ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1124,7 +1124,7 @@
     },
     {
       "id": 32116,
-      "content": "ɘ",
+      "content": "ɵ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1133,7 +1133,7 @@
     },
     {
       "id": 32117,
-      "content": "ɛ",
+      "content": "ɘ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1142,7 +1142,7 @@
     },
     {
       "id": 32118,
-      "content": "ɜ",
+      "content": "ɛ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1151,7 +1151,7 @@
     },
     {
       "id": 32119,
-      "content": "ɝ",
+      "content": "ɜ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1160,7 +1160,7 @@
     },
     {
       "id": 32120,
-      "content": "ɞ",
+      "content": "ɝ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1169,7 +1169,7 @@
     },
     {
       "id": 32121,
-      "content": "ɠ",
+      "content": "ɞ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1178,7 +1178,7 @@
     },
     {
       "id": 32122,
-      "content": "ɢ",
+      "content": "ɠ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1187,7 +1187,7 @@
     },
     {
       "id": 32123,
-      "content": "ʛ",
+      "content": "ɢ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1196,7 +1196,7 @@
     },
     {
       "id": 32124,
-      "content": "ɡ",
+      "content": "ʛ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1205,7 +1205,7 @@
     },
     {
       "id": 32125,
-      "content": "ħ",
+      "content": "ɡ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1214,7 +1214,7 @@
     },
     {
       "id": 32126,
-      "content": "ɦ",
+      "content": "ħ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1223,7 +1223,7 @@
     },
     {
       "id": 32127,
-      "content": "ɥ",
+      "content": "ɦ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1232,7 +1232,7 @@
     },
     {
       "id": 32128,
-      "content": "ɧ",
+      "content": "ɥ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1241,7 +1241,7 @@
     },
     {
       "id": 32129,
-      "content": "ʜ",
+      "content": "ɧ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1250,7 +1250,7 @@
     },
     {
       "id": 32130,
-      "content": "ɪ",
+      "content": "ʜ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1259,7 +1259,7 @@
     },
     {
       "id": 32131,
-      "content": "ɨ",
+      "content": "ɪ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1268,7 +1268,7 @@
     },
     {
       "id": 32132,
-      "content": "̈",
+      "content": "ɨ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1277,7 +1277,7 @@
     },
     {
       "id": 32133,
-      "content": "ʝ",
+      "content": "̈",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1286,7 +1286,7 @@
     },
     {
       "id": 32134,
-      "content": "ɟ",
+      "content": "ʝ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1295,7 +1295,7 @@
     },
     {
       "id": 32135,
-      "content": "ʄ",
+      "content": "ɟ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1304,7 +1304,7 @@
     },
     {
       "id": 32136,
-      "content": "ɫ",
+      "content": "ʄ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1313,7 +1313,7 @@
     },
     {
       "id": 32137,
-      "content": "ɭ",
+      "content": "ɫ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1322,7 +1322,7 @@
     },
     {
       "id": 32138,
-      "content": "ɬ",
+      "content": "ɭ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1331,7 +1331,7 @@
     },
     {
       "id": 32139,
-      "content": "ʟ",
+      "content": "ɬ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1340,7 +1340,7 @@
     },
     {
       "id": 32140,
-      "content": "ɮ",
+      "content": "ʟ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1349,7 +1349,7 @@
     },
     {
       "id": 32141,
-      "content": "ɱ",
+      "content": "ɮ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1358,7 +1358,7 @@
     },
     {
       "id": 32142,
-      "content": "ŋ",
+      "content": "ɱ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1367,7 +1367,7 @@
     },
     {
       "id": 32143,
-      "content": "ɲ",
+      "content": "ŋ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1376,7 +1376,7 @@
     },
     {
       "id": 32144,
-      "content": "ɳ",
+      "content": "ɲ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1385,7 +1385,7 @@
     },
     {
       "id": 32145,
-      "content": "ɴ",
+      "content": "ɳ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1394,7 +1394,7 @@
     },
     {
       "id": 32146,
-      "content": "ɔ",
+      "content": "ɴ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1403,7 +1403,7 @@
     },
     {
       "id": 32147,
-      "content": "œ",
+      "content": "ɔ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1412,7 +1412,7 @@
     },
     {
       "id": 32148,
-      "content": "ɒ",
+      "content": "œ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1421,7 +1421,7 @@
     },
     {
       "id": 32149,
-      "content": "ɶ",
+      "content": "ɒ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1430,7 +1430,7 @@
     },
     {
       "id": 32150,
-      "content": "ø",
+      "content": "ɶ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1439,7 +1439,7 @@
     },
     {
       "id": 32151,
-      "content": "ɸ",
+      "content": "ø",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1448,7 +1448,7 @@
     },
     {
       "id": 32152,
-      "content": "ɾ",
+      "content": "ɸ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1457,7 +1457,7 @@
     },
     {
       "id": 32153,
-      "content": "ɹ",
+      "content": "ɾ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1466,7 +1466,7 @@
     },
     {
       "id": 32154,
-      "content": "ʁ",
+      "content": "ɹ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1475,7 +1475,7 @@
     },
     {
       "id": 32155,
-      "content": "ʀ",
+      "content": "ʁ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1484,7 +1484,7 @@
     },
     {
       "id": 32156,
-      "content": "ɻ",
+      "content": "ʀ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1493,7 +1493,7 @@
     },
     {
       "id": 32157,
-      "content": "ɽ",
+      "content": "ɻ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1502,7 +1502,7 @@
     },
     {
       "id": 32158,
-      "content": "ɺ",
+      "content": "ɽ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1511,7 +1511,7 @@
     },
     {
       "id": 32159,
-      "content": "ʃ",
+      "content": "ɺ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1520,7 +1520,7 @@
     },
     {
       "id": 32160,
-      "content": "ʂ",
+      "content": "ʃ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1529,7 +1529,7 @@
     },
     {
       "id": 32161,
-      "content": "θ",
+      "content": "ʂ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1538,7 +1538,7 @@
     },
     {
       "id": 32162,
-      "content": "ʈ",
+      "content": "θ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1547,7 +1547,7 @@
     },
     {
       "id": 32163,
-      "content": "ʊ",
+      "content": "ʈ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1556,7 +1556,7 @@
     },
     {
       "id": 32164,
-      "content": "ʉ",
+      "content": "ʊ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1565,7 +1565,7 @@
     },
     {
       "id": 32165,
-      "content": "ʌ",
+      "content": "ʉ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1574,7 +1574,7 @@
     },
     {
       "id": 32166,
-      "content": "ʋ",
+      "content": "ʌ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1583,7 +1583,7 @@
     },
     {
       "id": 32167,
-      "content": "ⱱ",
+      "content": "ʋ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1592,7 +1592,7 @@
     },
     {
       "id": 32168,
-      "content": "ɯ",
+      "content": "ⱱ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1601,7 +1601,7 @@
     },
     {
       "id": 32169,
-      "content": "ʍ",
+      "content": "ɯ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1610,7 +1610,7 @@
     },
     {
       "id": 32170,
-      "content": "ɰ",
+      "content": "ʍ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1619,7 +1619,7 @@
     },
     {
       "id": 32171,
-      "content": "χ",
+      "content": "ɰ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1628,7 +1628,7 @@
     },
     {
       "id": 32172,
-      "content": "ɣ",
+      "content": "χ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1637,7 +1637,7 @@
     },
     {
       "id": 32173,
-      "content": "ʎ",
+      "content": "ɣ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1646,7 +1646,7 @@
     },
     {
       "id": 32174,
-      "content": "ʏ",
+      "content": "ʎ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1655,7 +1655,7 @@
     },
     {
       "id": 32175,
-      "content": "ɤ",
+      "content": "ʏ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1664,7 +1664,7 @@
     },
     {
       "id": 32176,
-      "content": "ʐ",
+      "content": "ɤ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1673,7 +1673,7 @@
     },
     {
       "id": 32177,
-      "content": "ʑ",
+      "content": "ʐ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1682,7 +1682,7 @@
     },
     {
       "id": 32178,
-      "content": "ʔ",
+      "content": "ʑ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1691,7 +1691,7 @@
     },
     {
       "id": 32179,
-      "content": "ʕ",
+      "content": "ʔ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1700,7 +1700,7 @@
     },
     {
       "id": 32180,
-      "content": "ʡ",
+      "content": "ʕ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1709,7 +1709,7 @@
     },
     {
       "id": 32181,
-      "content": "ʢ",
+      "content": "ʡ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1718,7 +1718,7 @@
     },
     {
       "id": 32182,
-      "content": "ˈ",
+      "content": "ʢ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1727,7 +1727,7 @@
     },
     {
       "id": 32183,
-      "content": "ˌ",
+      "content": "ˈ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1736,7 +1736,7 @@
     },
     {
       "id": 32184,
-      "content": "ː",
+      "content": "ˌ",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -1745,7 +1745,7 @@
     },
     {
       "id": 32185,
-      "content": "ʤ",
+      "content": "ː",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,

tokenizer_config.json

@@ -104,7 +104,7 @@
   "eos_token": "</s>",
   "extra_ids": 100,
   "model_max_length": 512,
-  "name_or_path": "mariopeng/phoneT5",
+  "name_or_path": "t5-small",
   "pad_token": "<pad>",
   "special_tokens_map_file": null,
   "tokenizer_class": "T5Tokenizer",