add tokenizer

.gitignore

@@ -0,0 +1 @@
+checkpoint-*/

runs/Jan29_20-22-30_69667d063c94/1643487828.9446776/events.out.tfevents.1643487828.69667d063c94.72.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b74aea50dd036a5f68b3b4b21725ce1ce91e94d6cc4996192d0704ed33516dc
+size 4555

runs/Jan29_20-22-30_69667d063c94/events.out.tfevents.1643487828.69667d063c94.72.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0fa4f48d44fccab37edfbe60979d776917cf3cef00dac8ebd43945f7858dfd7
+size 4463

vocab.json

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