chatsakura is a large multilingual conversational model based on Bloomz, supporting Chinese, English, Japanese, German, and French.
The model size is only 3B.
The current limitations of the model include:
- The model's ability in some scenarios involving mathematical reasoning, code, etc., still needs improvement.
- The model cannot identify harmful instructions.
Model precision | FP16 | int8 | int4 |
---|---|---|---|
Finetuned Model | chatSakura-3b | chatSakura-3b-int8 | chatSakura-3b-int4 |
Note: int4 and int8 precision must be run on a GPU and have not been tested on Windows yet.
conda install pytorch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1 pytorch-cuda=11.7 -c pytorch -c nvidia
pip install -r requirements.txt
python main.py
Model precision | FP16 | int8 | int4 |
---|---|---|---|
Finetuned Model | 10G | 6G | 4G(recommended 6G) |