add tokenizer

runs/Feb22_17-32-20_jake-a-01/events.out.tfevents.1645547608.jake-a-01.92577.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e74180911195610ddc6368c4a699dd2a4479596fc79bde6af2988cea6da3f9dc
-size 7732
+oid sha256:e88ae590a7ec99e67623d67417f0b4d8311c78e20c879a4d45a417fc611ff765
+size 8086

vocab.json

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