ymcui/Chinese-Mixtral
Chinese Mixtral is a sparse mixture-of-experts large language model adapted for Chinese via continual pretraining and instruction tuning.
Not currently ranked — collecting fresh signals.
star history
This project takes Mistral’s Mixtral model with sparse MoE architecture and performs Chinese-language continual pretraining on large-scale unlabeled data to create a Chinese base model. It further applies instruction tuning to produce the Chinese Mixtral-Instruct variant. The model natively supports up to 32K context with reported capability up to 128K, and improvements in math reasoning and code generation. Inference is supported via llama.cpp quantization, requiring as low as 16GB memory.
Frequently asked
- What is ymcui/Chinese-Mixtral?
- Chinese Mixtral is a sparse mixture-of-experts large language model adapted for Chinese via continual pretraining and instruction tuning.
- Is Chinese-Mixtral open source?
- Yes — ymcui/Chinese-Mixtral is open source, released under the Apache-2.0 license.
- What language is Chinese-Mixtral written in?
- ymcui/Chinese-Mixtral is primarily written in Python.
- How popular is Chinese-Mixtral?
- ymcui/Chinese-Mixtral has 612 stars on GitHub.
- Where can I find Chinese-Mixtral?
- ymcui/Chinese-Mixtral is on GitHub at https://github.com/ymcui/Chinese-Mixtral.