fix(tokenizer): Sets fim and think tokens to non-special, and removes unk_token.

tokenizer.json

@@ -3,15 +3,6 @@
   "truncation": null,
   "padding": null,
   "added_tokens": [
-    {
-      "id": 5809,
-      "content": "�",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": false,
-      "special": true
-    },
     {
       "id": 100256,
       "content": "<|dummy_0|>",
@@ -37,7 +28,7 @@
       "lstrip": true,
       "rstrip": true,
       "normalized": false,
-      "special": true
+      "special": false
     },
     {
       "id": 100259,
@@ -46,7 +37,7 @@
       "lstrip": true,
       "rstrip": true,
       "normalized": false,
-      "special": true
+      "special": false
     },
     {
       "id": 100260,
@@ -55,7 +46,7 @@
       "lstrip": true,
       "rstrip": true,
       "normalized": false,
-      "special": true
+      "special": false
     },
     {
       "id": 100261,
@@ -865,7 +856,7 @@
       "lstrip": true,
       "rstrip": true,
       "normalized": false,
-      "special": true
+      "special": false
     },
     {
       "id": 100351,
@@ -874,7 +865,7 @@
       "lstrip": true,
       "rstrip": true,
       "normalized": false,
-      "special": true
+      "special": false
     }
   ],
   "normalizer": null,