mistralai/Mistral-7B-v0.1 · Lora fine tuning for text classification with Peft

I am trying to fine tune mistralai/Mistral-7B-v0.1 for text classification with 3 labels. my dataset is rather small.
I want to use Peft for Low-Rank Adaptation, However, it seems it can not find the labels in my dataset since it throws :

ValueError: The model did not return a loss from the inputs, only the following keys: last_hidden_state,pooler_output. For reference, the inputs it received are input_ids,token_type_ids,attention_mask.

I was wondering if anyone had the same experience or knows of any solution.