This repo contains the core implementation for InspectCoder and InspectWare. To run the script, please prepare an external module LLM for LLM inference, put the module paths to config.yml. This will be used in from request_model import LLM; llm = LLM(model_name=args.model_name) in InspectCoder. To prepare the execution environments, please also download the LiveCodeBench and BigCodeBench repositories and put their paths to config.yml.
We also provide full logs of case studies conducted in our paper at case_study for your convenient review.