fix: Correct tokenizer configuration for HF

hf_model/tokenizer_config.json

@@ -1,5 +1,4 @@
 {
   "model_type": "gpt2",
-  "tokenizer_class": "T5Tokenizer",
-  "sp_model_file": "spm_chat.model"
-}
+  "tokenizer_class": "T5Tokenizer"
+}