This repo contains a low-rank adapter for LLaMA-7b trained on Bulgarian dataset.
This model was introduced in
this paper.
The training data is private Bulgarian dataset.
This is an instruction-based model similart to ChatGPT but in Bulgarian.
1>>> git clone https://github.com/tloen/alpaca-lora.git
2>>> cd alpaca-lora
3>>> pip install -r requirements.txt
4>>>
5>>> python generate.py \
6 --load_8bit \
7 --base_model 'yahma/llama-7b-hf' \
8 --lora_weights 'rmihaylov/alpaca-lora-bg-7b' \
9 --share_gradio
This will download both a base model and an adapter from huggingface. Then it will run a gradio interface for chatting.
Example using this model:
Colab. You need a Colab Pro because the model needs high ram when loading.
