OpenLLaMA-Chinese is a 100% free Chinese large language model, and can be utilized for both non-commercial and commercial purposes.
OpenLLaMA-Chinese is built on OpenLLaMA, which is a permissively licensed open-source reproduction of Meta AI's LLaMA 7B and 13B models, trained on the RedPajama dataset. OpenLLaMA also includes a smaller 3B variant of the LLaMA model. We have conducted fine-tuning on Chinese and English instructions using the OpenLLaMA base models and have made our weights publicly available.
[2023/06/29] We released the openllama 13b model by using evol intructions.
[2023/06/24] We use evol intructions from WizardLM to finetune the openllama 7B, the 13B Model will be avaliable in next week!.
For Chinese fine-tuning, we utilized the alpaca_data_zh_51k.json from the Chinese-LLaMA-Alpaca project.
For English fine-tuning, we employed the alpaca_data.json from the StanfordAlpaca project.
For fine-tuning with both English and Chinese instructions, we used data from both sources.
We modified the generate code from LLaMA-X.
To use the PyTorch inference code, follow these steps:
- Download the weights and update the base_model path in inference/gen_torch.py.
- Run the following command:
python inference/gen_torch.pyFittenTech offers LLMs pretraining and fine-tuning services. For more details, please visit https://llm.fittentech.com/.
We would like to express our gratitude to the developers of the following open-source projects, as our project builds upon their work:
We adopt the Apache License, following OpenLLaMA's license.
