About the projects: We are building LLM evaluation and training datasets to train LLM to work on realistic software engineering problems. One of our approaches, in this project, is to build verifiable SWE tasks based on public repository histories in...