Upload tokenizer

special_tokens_map.json

@@ -1,13 +1,13 @@
 {
   "bos_token": {
-    "content": "<|startoftext|>",
+    "content": "<s>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,
     "single_word": false
   },
   "eos_token": {
-    "content": "<|endoftext|>",
+    "content": "</s>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,

tokenizer.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:386c49cf943d71aa110361135338c50e38beeff0a66593480421f37b319e1a39
-size 1033105
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

tokenizer_config.json

@@ -1,5 +1,5 @@
 {
-  "add_bos_token": false,
+  "add_bos_token": true,
   "add_eos_token": false,
   "added_tokens_decoder": {
     "0": {
@@ -11,7 +11,7 @@
       "special": true
     },
     "1": {
-      "content": "<|startoftext|>",
+      "content": "<s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -19,7 +19,7 @@
       "special": true
     },
     "2": {
-      "content": "<|endoftext|>",
+      "content": "</s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -28,11 +28,11 @@
     }
   },
   "additional_special_tokens": [],
-  "bos_token": "<|startoftext|>",
+  "bos_token": "<s>",
   "chat_template": null,
   "clean_up_tokenization_spaces": false,
   "cls_token": null,
-  "eos_token": "<|endoftext|>",
+  "eos_token": "</s>",
   "mask_token": null,
   "model_input_names": [
     "input_ids",