add tokenizer

runs/Jun19_15-03-40_container-46581190ae-2377fc8a/1655624293.6334171/events.out.tfevents.1655624293.container-46581190ae-2377fc8a.19321.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:602b1695b51bda1c1594de97ec6da6f9b5874b075d5e80ff2638e2568ddb52f9
+size 5380

runs/Jun19_15-03-40_container-46581190ae-2377fc8a/events.out.tfevents.1655624293.container-46581190ae-2377fc8a.19321.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db9ccb383ef843e838596bdd3e7c449d49325de21658fc06ba614e47a8de4b62
+size 5492

vocab.json

@@ -1,37 +1,37 @@
 {
-  "'": 3,
-  "2": 31,
-  "3": 30,
-  "4": 24,
-  "5": 2,
-  "6": 5,
+  "'": 27,
+  "2": 21,
+  "3": 15,
+  "4": 14,
+  "5": 17,
+  "6": 19,
   "[PAD]": 34,
   "[UNK]": 33,
-  "a": 21,
-  "b": 12,
-  "c": 15,
-  "d": 14,
-  "e": 10,
-  "f": 9,
-  "g": 13,
-  "h": 8,
-  "i": 18,
-  "j": 17,
-  "k": 22,
-  "l": 0,
-  "m": 20,
-  "n": 11,
-  "o": 4,
-  "p": 1,
-  "q": 26,
-  "r": 7,
-  "s": 19,
-  "t": 32,
-  "u": 28,
-  "v": 25,
-  "w": 16,
-  "x": 6,
-  "y": 23,
-  "z": 29,
-  "|": 27
+  "a": 22,
+  "b": 32,
+  "c": 18,
+  "d": 26,
+  "e": 31,
+  "f": 7,
+  "g": 0,
+  "h": 6,
+  "i": 23,
+  "j": 3,
+  "k": 10,
+  "l": 16,
+  "m": 11,
+  "n": 5,
+  "o": 24,
+  "p": 29,
+  "q": 25,
+  "r": 2,
+  "s": 13,
+  "t": 9,
+  "u": 20,
+  "v": 12,
+  "w": 28,
+  "x": 4,
+  "y": 8,
+  "z": 30,
+  "|": 1
 }