--- library_name: transformers tags: - DPO license: apache-2.0 datasets: - lightblue/response-dataset-plus-qwen-judged language: - ja base_model: - Qwen/Qwen2.5-7B-Instruct --- [日本語モデルカード/Japanese model card](#japanese) [日本語のブログ/Full Japanese dev blog](https://note.com/lightblue_tech/n/n6967ff462f4a?sub_rt=share_pb) [Development source code/開発ソースコード](https://github.com/lightblue-tech/karasu_dpo_202501) # Karasu-DPO-7B This is a Japanese version of the [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) model which was DPO trained using synthetic Japanese conversation data. This model outperforms the base [Qwen/Qwen2.5-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-7B-Instruct) model on the [arena-hard-auto-multilingual](https://github.com/lightblue-tech/arena-hard-auto-multilingual) chat benchmark: |Qwen2.5-7B-Instruct|Karasu-DPO-7B| |----|----| |50.0|66.2| We recommend this model for use as a general conversation AI. # How to use This model can be used in the same way as any Qwen 2.5 model. We recommend using vLLM for simplicity and speed.