[Project] Zicklein - a German 🇩🇪 fine-tuned LlaMA-7b base model (OS)

Zicklein is a German version of Alpaca 7b fine-tuned using the LoRA method, trained using a German translated version of the cleaned Alpaca instruct dataset.

Github: [https://github.com/avocardio/zicklein](https://github.com/avocardio/zicklein)

HuggingFace: [https://huggingface.co/avocardio/alpaca-lora-7b-german-base-52k](https://huggingface.co/avocardio/alpaca-lora-7b-german-base-52k)

You can also try it out [here](https://huggingface.co/spaces/avocardio/German-Alpaca-LoRA-7b) (although its super slow - running on a CPU, responses take around 130s).

alpaca dataset german llama lora machinelearning project translated