add tokenizer

added_tokens.json

@@ -1,4 +1,4 @@
 {
-  "</s>": 30,
-  "<s>": 29
+  "</s>": 29,
+  "<s>": 28
 }

runs/Jul21_15-05-08_Tejas-XPS22/1658430356.7334583/events.out.tfevents.1658430356.Tejas-XPS22.1656.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6effd257ea22f1c6637a0fc79ee01673fcc008dc5d43eba29a96440710629916
+size 5385

runs/Jul21_15-05-08_Tejas-XPS22/events.out.tfevents.1658430356.Tejas-XPS22.1656.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1a6972ea857ef8e2fecbe511b8a20a4b000442d7e75ed44356b07aebf3389b5
+size 5092

special_tokens_map.json

@@ -18,5 +18,5 @@
   "bos_token": "<s>",
   "eos_token": "</s>",
   "pad_token": "[PAD]",
-  "unk_token": "[UNK]"
+  "unk_token": "<unk>"
 }

tokenizer_config.json

@@ -7,6 +7,6 @@
   "replace_word_delimiter_char": " ",
   "special_tokens_map_file": null,
   "tokenizer_class": "Wav2Vec2CTCTokenizer",
-  "unk_token": "[UNK]",
+  "unk_token": "<unk>",
   "word_delimiter_token": " "
 }

vocab.json

@@ -1,7 +1,6 @@
 {
   " ": 0,
-  "[PAD]": 28,
-  "[UNK]": 27,
+  "[PAD]": 27,
   "a": 1,
   "b": 2,
   "c": 3,