Local assistant, Mike, is an AI assistant chatbot designed to run in your personal PC. Mike can help you to increase your productivity without any risk on your privacy. Upload your files for answers grounded on your documents. For general knowledge and up-to-date information, the assistant may make use of web search. For specialized coding tasks the assistant may make use of a code llm.
Local assistant relies on Ollama to host a local language model. You need to install Ollama and download a language model.
Go to Ollama Home Page. Choose your platform to install. And follow the instructions.
Download a language model. Llama3 is recommended:
ollama pull llama3Download a code model. eramax/nxcode-cq-7b-orpo:q6 is recommended:
ollama pull eramax/nxcode-cq-7b-orpo:q6Download an embedding model.
ollama pull nomic-embed-textPlease note that language models run best at GPUs. If you do not have a GPU you can still use a language model, but it will be slow. A typical language model will require approximately 5GB of disk space.
For a list of available models you may refer to Ollama Models Library
Go to Official Python Downloads Page.
Use install scripts provided for your platform e.g. windows, linux.
For linux environment you may need to give execute permission to the scripts.
chmod +x install-linux.shConfiguration file is config/config.yml. Edit it to configure options like models, database, etc.
Prompts are config/prompts.yml file. You may edit the file for providing your own prompts.
Use launch script provided for your platform e.g. windows, linux.
For linux environment you may need to give execute permission to the scripts.
chmod +x launch-linux.shOpen chat screen at Assistant Chat.
You may upload your files using the upload screen at Upload Page.
If you encounter an issue feel free to report by opening a GitHub issue.