Adding Tests

[BassCoder2808]

• Currently there tests such as unit tests or something similar to that for testing the various use cases of the application
• Addition of inbuilt streamlit-test would be beneficial for checking the various features available.

Would be happy to contribute a PR for the same

[frog-land]

There is very little front end functionality in the Streamlit Chat2VIS app. It basically acts as a shell to demonstrate the use of LLMs to generate data visualisations. By unit testing are you referring to unit testing of the Python code? If so, then no, there are no unit test scripts. If you are referring to unit testing of the research, then you can find what we tested by reading the articles https://doi.org/10.1109/ACCESS.2023.3274199 and https://arxiv.org/abs/2303.14292 (the Code Llama one is still under journal review and yet to be published). The versions of the LLM models we originally tested them on are up to a year old now, and it would be interesting how their behaviour has changed. I have even found discussions about deterioration in code generation of chatgpt based models. Anyway, you are most welcome to contribute to testing of the Python code if that is of value to you. It wasn’t the focus of the research but I am happy to consider incorporating fixes to what you find. Thanks for your interest and making contact. It’s greatly appreciated.