eliseobao commited on
Commit
c1c2d64
·
1 Parent(s): 7ad5799

Align tokenizer_config chat template with docs.mistral.ai chat template

Browse files

As others mentioned, there is a subtle difference between chat templates:
[https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1/discussions/53](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1/discussions/53)

This edit aligns with:
[https://docs.mistral.ai/llm/mistral-instruct-v0.1#chat-template](https://docs.mistral.ai/llm/mistral-instruct-v0.1#chat-template)
[https://docs.mistral.ai/usage/guardrailing#appendix](https://docs.mistral.ai/usage/guardrailing#appendix)

If accepted, model card should be modified accordingly.

Files changed (1) hide show
  1. tokenizer_config.json +1 -1
tokenizer_config.json CHANGED
@@ -29,7 +29,7 @@
29
  },
30
  "additional_special_tokens": [],
31
  "bos_token": "<s>",
32
- "chat_template": "{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token + ' ' }}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}",
33
  "clean_up_tokenization_spaces": false,
34
  "eos_token": "</s>",
35
  "legacy": true,
 
29
  },
30
  "additional_special_tokens": [],
31
  "bos_token": "<s>",
32
+ "chat_template": "{{ bos_token }}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if message['role'] == 'user' %}{{ '[INST] ' + message['content'] + ' [/INST] ' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token }}{% else %}{{ raise_exception('Only user and assistant roles are supported!') }}{% endif %}{% endfor %}",
33
  "clean_up_tokenization_spaces": false,
34
  "eos_token": "</s>",
35
  "legacy": true,