This package was build within SemANT project https://semant.cz/. It contains tools for communicating with LLMs via commercial API and performing RAG-based task on collection of documents.
You must install the faiss (faiss~=1.7.3):
conda install -c pytorch faiss-gpu
or
conda install -c conda-forge faiss-gpu
To install required packages, run:
pip install -r requirements.txt
To use this tool you need a valid OpenAI API key. Get your key and fill it in appropriate config file. There are several already prepared use cases.
You can find their configuration files in use_cases directory. Read their prompts to get more information about their usage.
Also, you need to have at least one semantindexer collection (https://github.com/DCGM/semantindexer) of documents in collections directory (it can be changed in a config).
To use LLM for your task choose one config and collection of documents and run:
./run.py ask name_of_collection_folder -c path_to_config