add tokenizer

runs/May09_10-46-50_hydcorp-esxi09/1652073488.8618567/events.out.tfevents.1652073488.hydcorp-esxi09.1933037.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:88eb93ffa799e79046b7e0e45ec10c2cf175dc0c0e7c52e1315f13c999dc43a7
+size 4482

runs/May09_10-46-50_hydcorp-esxi09/events.out.tfevents.1652073488.hydcorp-esxi09.1933037.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fbc917111323b15d0041d992ad487a7d844ddaa1710a5817aa4ffdb2ceedc601
+size 4145

vocab.json

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