Deci

/

DeciLM-7B-instruct

like 96

Thanks for this release.

I am opening this discussion because I know the prompt format is like following:

### System:
{system}
### User:
{usr}
### Assistant:

But it would be fantastic if you can add the chat template to the tokenizer_config.json as many other models do. Here is an example what I mean. This makes easier integrate the model because you are ensuring it is correctly prompted by just a single line of code: tokenizer.apply_chat_template()

Thank you so much again.