Add `chat_template` in tokenizer config

As dolphin-mistral follows chatml template, adding `chat_template` in tokenizer config can be helpful to users.

Also text-generation-inference has supported chat_template since 1.4.0

tokenizer_config.json

@@ -37,6 +37,7 @@
   },
   "additional_special_tokens": [],
   "bos_token": "<s>",
+  "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|im_start|>user\n' + message['content'] + '<|im_end|>' }}\n{% elif message['role'] == 'system' %}\n{{ '<|im_start|>system\n' + message['content'] + '<|im_end|>' }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|im_start|>assistant\n' + message['content'] + '<|im_end|>' }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|im_start|>assistant' }}\n{% endif %}\n{% endfor %}",
   "clean_up_tokenization_spaces": false,
   "eos_token": "<|im_end|>",
   "legacy": true,