Static Code Analaysis for LLM's
Short experiment run by me and my peers for High Performance Computing. If the reader is interested in the methodology or experiment, look for Report_LLM_Code_genration_Analysis.pdf in the repo.
Crux of the study:
We task different LLM's to generate code on 5 standard pre-decided problems, then run the solutions through Pylint. Some very interesting results that we found.
The numbers present the Pylint score on each kind of problem (rows) for different LLM's (column)