jhflow/mistral7b-lora-multi-turn-v2

This is finetuned model for multi turn style RAG. I picked up some datasets consisting of knowledge based multi turn conversations for training.

base model : https://huggingface.co/Open-Orca/Mistral-7B-OpenOrca
prompt teamplate : chatml (same as https://huggingface.co/Open-Orca/Mistral-7B-OpenOrca)

<|im_start|>system
{system_prompt}

{context}<|im_end|>
<|im_start|>user
{user_query}<|im_end|>
<|im_start|>assistant
{answer}
<|im_end|>