add tokenizer

runs/Nov05_09-17-33_29f01e962d6c/1636103870.8574052/events.out.tfevents.1636103870.29f01e962d6c.78.5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fa1f2f8e07eaa5d39a334784067356b34643adb463a0b938172d2a84c2c43aaf
+size 4576

runs/Nov05_09-17-33_29f01e962d6c/events.out.tfevents.1636103870.29f01e962d6c.78.4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c557d8dd07751bf8f685557c97c10ed1b1fde5de78dea2b0a7728acb5205580b
+size 4476

vocab.json

@@ -1 +1 @@
-{"ّ": 0, "د": 1, "ه": 2, "ن": 3, "ط": 4, "ي": 5, "ف": 6, "ذ": 7, "ٌ": 8, "ْ": 9, "خ": 10, "ش": 11, "ؤ": 12, "غ": 13, "َ": 14, "ك": 15, "ص": 16, "ح": 17, "ئ": 19, "ُ": 20, "ٍ": 21, "ب": 22, "م": 23, "؟": 24, "ى": 25, "ظ": 26, "ل": 27, "ِ": 28, "أ": 29, "ج": 30, "ع": 31, "ً": 32, "ا": 33, "،": 34, "إ": 35, "و": 36, "ء": 37, "ق": 38, "س": 39, "ز": 40, "ت": 41, "ض": 42, "ة": 43, "ث": 44, "ر": 45, "آ": 46, "|": 18, "[UNK]": 47, "[PAD]": 48}
+{"أ": 0, "ب": 1, "ظ": 2, "ث": 3, "ذ": 4, "؟": 5, "ي": 6, "ّ": 7, "ئ": 8, "،": 9, "ل": 10, "ز": 11, "ة": 12, "ح": 13, "ع": 14, "ض": 15, "ج": 16, "ن": 17, "َ": 18, "ق": 19, "ش": 20, "ص": 21, "ه": 22, "ر": 23, "إ": 24, "د": 25, "ت": 26, "خ": 27, "ً": 29, "ك": 30, "ؤ": 31, "م": 32, "ِ": 33, "ٍ": 34, "ء": 35, "ٌ": 36, "و": 37, "س": 38, "ُ": 39, "آ": 40, "ا": 41, "ط": 42, "ى": 43, "ف": 44, "ْ": 45, "غ": 46, "|": 28, "[UNK]": 47, "[PAD]": 48}