Training in progress, step 20000

config.json

@@ -27,5 +27,5 @@
   "torch_dtype": "float32",
   "transformers_version": "4.44.2",
   "use_cache": true,
-  "vocab_size": 31
+  "vocab_size": 115
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a555fb95a12d45a694e96ce59d9cf43d69a3e5476b5d3b277ab7f23d8c179c5
-size 19257664
+oid sha256:42c3ecb6fd8d551111138ba77f01ed049a5b5cb50d1924211d6773c7e249584c
+size 19343680

tokenizer.json

@@ -51,7 +51,7 @@
     ]
   },
   "pre_tokenizer": {
-    "type": "Whitespace"
+    "type": "WhitespaceSplit"
   },
   "post_processor": {
     "type": "TemplateProcessing",
@@ -115,33 +115,117 @@
       "PAD": 1,
       "WORD_BOUNDARY": 2,
       "UTT_BOUNDARY": 3,
-      "a": 4,
-      "i": 5,
-      "ɾ": 6,
-      "e̞": 7,
+      "a˧˥": 4,
+      "u˧˥": 5,
+      "a˥": 6,
+      "au": 7,
       "n": 8,
-      "k": 9,
-      "ɲ": 10,
-      "o̞": 11,
-      "m": 12,
-      "s": 13,
-      "u": 14,
+      "a˥˩": 9,
+      "ʃ̺": 10,
+      "ɻ̩˥˩": 11,
+      "ə˧˥": 12,
+      "m": 13,
+      "ɤ": 14,
       "p": 15,
-      "d": 16,
-      "l": 17,
-      "t": 18,
-      "β": 19,
-      "ɡ": 20,
+      "j": 16,
+      "e˧˥": 17,
+      "kʰ": 18,
+      "k": 19,
+      "ɤ˥˩": 20,
       "w": 21,
-      "ʝ": 22,
-      "f": 23,
-      "x": 24,
-      "j": 25,
-      "r": 26,
-      "t̠ʃ": 27,
-      "ʃ": 28,
-      "tl": 29,
-      "ts": 30
+      "o˥": 22,
+      "t̠ʃ̺ʰ": 23,
+      "ə˥": 24,
+      "ŋ": 25,
+      "t": 26,
+      "ʊ˥": 27,
+      "ɕ": 28,
+      "i": 29,
+      "a": 30,
+      "l": 31,
+      "au˧˩˧": 32,
+      "x": 33,
+      "u˧˩˧": 34,
+      "i˥": 35,
+      "ei˧˩˧": 36,
+      "pʰ": 37,
+      "i˧˥": 38,
+      "ai˧˥": 39,
+      "ou˧˩˧": 40,
+      "ɤ˧˥": 41,
+      "o˧˩˧": 42,
+      "tɕ": 43,
+      "au˥˩": 44,
+      "ts": 45,
+      "ə˧˩˧": 46,
+      "ɤ˥": 47,
+      "ei˧˥": 48,
+      "ʊ˧˥": 49,
+      "i˧˩˧": 50,
+      "t̠ʃ̺": 51,
+      "ɻ̩˧˩˧": 52,
+      "ei˥˩": 53,
+      "s": 54,
+      "u˥˩": 55,
+      "ɹ̪̩": 56,
+      "ai˥": 57,
+      "u˥": 58,
+      "tɕʰ": 59,
+      "a˧˩˧": 60,
+      "ai˥˩": 61,
+      "ɛ˥˩": 62,
+      "f": 63,
+      "i˥˩": 64,
+      "y˥˩": 65,
+      "au˧˥": 66,
+      "ɻ": 67,
+      "ou˥˩": 68,
+      "e˥": 69,
+      "tʰ": 70,
+      "ɹ̪̩˥˩": 71,
+      "ɛ˧˥": 72,
+      "au˥": 73,
+      "ou˧˥": 74,
+      "e˧˩˧": 75,
+      "ɛ˥": 76,
+      "ɻ̩˥": 77,
+      "ɥ": 78,
+      "ɹ̪̩˧˩˧": 79,
+      "ai˧˩˧": 80,
+      "ou˥": 81,
+      "o˥˩": 82,
+      "ɛ˧˩˧": 83,
+      "ʊ˧˩˧": 84,
+      "ɔ˥": 85,
+      "tsʰ": 86,
+      "ei": 87,
+      "ə˥˩": 88,
+      "o": 89,
+      "ʊ˥˩": 90,
+      "ou": 91,
+      "ɤ˧˩˧": 92,
+      "o˧˥": 93,
+      "ei˥": 94,
+      "e˥˩": 95,
+      "ɚ˧˩˧": 96,
+      "y˥": 97,
+      "ɚ˥˩": 98,
+      "y˧˥": 99,
+      "ɻ̩": 100,
+      "y˧˩˧": 101,
+      "ɹ̪̩˥": 102,
+      "ɻ̩˧˥": 103,
+      "u": 104,
+      "ə": 105,
+      "ai": 106,
+      "ʊ": 107,
+      "e": 108,
+      "ɚ˧˥": 109,
+      "ɔ˥˩": 110,
+      "ɹ̪̩˧˥": 111,
+      "ɛ": 112,
+      "y": 113,
+      "m˧˥": 114
     },
     "unk_token": "UNK"
   }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:82b3efe1c353bf704a32a56d3e2d90a521ee074a5be0597867a8e435f00dc496
+oid sha256:084e0397b40c5e56e7d3df631100c29dc2e74d25bb673e1905a6ce4ebdfd3f52
 size 5368

vocab.json

@@ -1 +1 @@
-{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"a":4,"i":5,"ɾ":6,"e̞":7,"n":8,"k":9,"ɲ":10,"o̞":11,"m":12,"s":13,"u":14,"p":15,"d":16,"l":17,"t":18,"β":19,"ɡ":20,"w":21,"ʝ":22,"f":23,"x":24,"j":25,"r":26,"t̠ʃ":27,"ʃ":28,"tl":29,"ts":30}
+{"UNK":0,"PAD":1,"WORD_BOUNDARY":2,"UTT_BOUNDARY":3,"a˧˥":4,"u˧˥":5,"a˥":6,"au":7,"n":8,"a˥˩":9,"ʃ̺":10,"ɻ̩˥˩":11,"ə˧˥":12,"m":13,"ɤ":14,"p":15,"j":16,"e˧˥":17,"kʰ":18,"k":19,"ɤ˥˩":20,"w":21,"o˥":22,"t̠ʃ̺ʰ":23,"ə˥":24,"ŋ":25,"t":26,"ʊ˥":27,"ɕ":28,"i":29,"a":30,"l":31,"au˧˩˧":32,"x":33,"u˧˩˧":34,"i˥":35,"ei˧˩˧":36,"pʰ":37,"i˧˥":38,"ai˧˥":39,"ou˧˩˧":40,"ɤ˧˥":41,"o˧˩˧":42,"tɕ":43,"au˥˩":44,"ts":45,"ə˧˩˧":46,"ɤ˥":47,"ei˧˥":48,"ʊ˧˥":49,"i˧˩˧":50,"t̠ʃ̺":51,"ɻ̩˧˩˧":52,"ei˥˩":53,"s":54,"u˥˩":55,"ɹ̪̩":56,"ai˥":57,"u˥":58,"tɕʰ":59,"a˧˩˧":60,"ai˥˩":61,"ɛ˥˩":62,"f":63,"i˥˩":64,"y˥˩":65,"au˧˥":66,"ɻ":67,"ou˥˩":68,"e˥":69,"tʰ":70,"ɹ̪̩˥˩":71,"ɛ˧˥":72,"au˥":73,"ou˧˥":74,"e˧˩˧":75,"ɛ˥":76,"ɻ̩˥":77,"ɥ":78,"ɹ̪̩˧˩˧":79,"ai˧˩˧":80,"ou˥":81,"o˥˩":82,"ɛ˧˩˧":83,"ʊ˧˩˧":84,"ɔ˥":85,"tsʰ":86,"ei":87,"ə˥˩":88,"o":89,"ʊ˥˩":90,"ou":91,"ɤ˧˩˧":92,"o˧˥":93,"ei˥":94,"e˥˩":95,"ɚ˧˩˧":96,"y˥":97,"ɚ˥˩":98,"y˧˥":99,"ɻ̩":100,"y˧˩˧":101,"ɹ̪̩˥":102,"ɻ̩˧˥":103,"u":104,"ə":105,"ai":106,"ʊ":107,"e":108,"ɚ˧˥":109,"ɔ˥˩":110,"ɹ̪̩˧˥":111,"ɛ":112,"y":113,"m˧˥":114}