Trying to use various automatic tools and LLMs to speed up development, debugging and optimization of number-crunching code for scientific calculations
tests- containg something like payground for experimenting, testing and examples of different LLM applications and tasks. It is meant for development.scripts- containin finished scripts for different taskspyCruncher- the main python library for automatization of using LLMs for different tasks such as summarization, code generation, etc.
- On Ubuntu linux it is recommended to use python envirnoment which allows to install python packages with
pip. For exampplealias activate_ml="source ~/venvs/ML/bin/activate" activate_ml - Then simply navigate to
scriptsortestsand run the relevant script
Make sure to install all dependencies using pip install -r requirements.txt
- Key dependencies are:
jsontomlopenai
- Additional dependencies ( required by some modules) are:
google-generativeaitree-sitterMaxima