wangrongsheng / aurora

šŸ³ Aurora represents the Chinese version of the MoE model, refined from the Mixtral-8x7B architecture. It adeptly unlocks the modelā€™s potential for bilingual dialogue in both Chinese and English across a wide range of open-domain topics.

8x7B

62 Pulls Updated 6 months ago

6 months ago

0a77bc646fee Ā· 26GB

model
llama
Ā·
46.7B
Ā·
Q4_0
template
<|im_start|>system {{ .System }}<|im_end|> <|im_start|>user {{ .Prompt }}<|im_end|> <|im_start|>assistant
params
{"stop":["<|im_start|>","<|im_end|>"]}
system
You are Aurora, a helpful AI assistant.

Readme

Overview

Aurora represents the Chinese iteration of the MoE model, refined from the Mixtral-8x7B architecture. It adeptly unlocks the modelā€™s potential for bilingual dialogue in both Chinese and English across a wide range of open-domain topics. Our Github is https://github.com/WangRongsheng/Aurora!

Run Aurora model

ollama run wangrongsheng/aurora

# or
# single-round dialogue
ollama run wangrongsheng/aurora "What is your favourite condiment?"

Check Aurora model

ollama ls

wangrongsheng/aurora will be listed with other models.

Remove Aurora model

ollama rm wangrongsheng/aurora

Use by API

curl -X POST http://localhost:11434/api/generate -d '{
  "model": "wangrongsheng/aurora",
  "prompt":"What is your favourite condiment?"
 }'

Citation

If you find our work helpful, feel free to give us a cite.

@misc{wang2023auroraactivating,
      title={Aurora:Activating Chinese chat capability for Mixtral-8x7B sparse Mixture-of-Experts through Instruction-Tuning}, 
      author={Rongsheng Wang and Haoming Chen and Ruizhe Zhou and Yaofei Duan and Kunyan Cai and Han Ma and Jiaxi Cui and Jian Li and Patrick Cheong-Iao Pang and Yapeng Wang and Tao Tan},
      year={2023},
      eprint={2312.14557},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}