The models trained with Llama, and Mistral on FewRel dataset. The paper of the work has been submitted to possible journal