add tokenizer

runs/Jun18_15-32-00_container-3d5811a0ae-620c5fad/1655539544.9155424/events.out.tfevents.1655539544.container-3d5811a0ae-620c5fad.62031.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61e8370ce845cb7ecdc6a7caf4351891296b5f6eb5194597a82bed2b7a6e870d
+size 5380

runs/Jun18_15-32-00_container-3d5811a0ae-620c5fad/events.out.tfevents.1655539544.container-3d5811a0ae-620c5fad.62031.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26dc510c029806e7ac87aaa52fed6e2ddd26758e1a0ea04a95242ad2e3a75ccc
+size 5335

vocab.json

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