add tokenizer

runs/Mar02_11-41-46_9e85cc13cb13/1677757311.048026/events.out.tfevents.1677757311.9e85cc13cb13.2668.19

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c7fe4e837c88f934f7a280b66c04d612740c0b496a5206254db15632d1dfcd4
+size 5385

runs/Mar02_11-41-46_9e85cc13cb13/events.out.tfevents.1677757311.9e85cc13cb13.2668.18

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c37f283b277af7d0b7f9ad60df737c732306a8936a52c5a4688c0f78e46e2eb9
+size 4136

vocab.json

@@ -1,47 +1,47 @@
 {
-  " ": 33,
+  " ": 13,
   "PAD": 43,
   "[UNK]": 44,
-  "آ": 1,
-  "أ": 42,
-  "ؤ": 21,
-  "ئ": 28,
-  "ا": 9,
-  "ب": 34,
-  "ت": 22,
-  "ث": 37,
-  "ج": 30,
-  "ح": 35,
-  "خ": 19,
-  "د": 7,
-  "ذ": 29,
-  "ر": 0,
-  "ز": 18,
-  "س": 39,
-  "ش": 20,
-  "ص": 13,
-  "ض": 26,
-  "ط": 11,
-  "ظ": 40,
-  "ع": 23,
-  "غ": 8,
-  "ف": 5,
-  "ق": 6,
-  "ل": 12,
-  "م": 3,
-  "ن": 27,
-  "و": 41,
-  "ً": 14,
-  "ٹ": 17,
-  "پ": 25,
-  "چ": 24,
-  "ڈ": 32,
-  "ڑ": 10,
-  "ک": 36,
-  "گ": 16,
-  "ں": 15,
-  "ھ": 38,
-  "ہ": 2,
-  "ی": 31,
-  "ے": 4
+  "آ": 24,
+  "أ": 40,
+  "ؤ": 32,
+  "ئ": 29,
+  "ا": 3,
+  "ب": 37,
+  "ت": 14,
+  "ث": 34,
+  "ج": 20,
+  "ح": 9,
+  "خ": 26,
+  "د": 36,
+  "ذ": 41,
+  "ر": 35,
+  "ز": 19,
+  "س": 5,
+  "ش": 23,
+  "ص": 8,
+  "ض": 27,
+  "ط": 0,
+  "ظ": 2,
+  "ع": 42,
+  "غ": 10,
+  "ف": 25,
+  "ق": 28,
+  "ل": 11,
+  "م": 4,
+  "ن": 7,
+  "و": 1,
+  "ً": 39,
+  "ٹ": 12,
+  "پ": 6,
+  "چ": 15,
+  "ڈ": 21,
+  "ڑ": 17,
+  "ک": 38,
+  "گ": 22,
+  "ں": 16,
+  "ھ": 31,
+  "ہ": 18,
+  "ی": 30,
+  "ے": 33
 }