tasnim7ahmed/simcode-bench ? reverse-engineered prompt

Build me a project that acts as a benchmark for turning plain English networking problems into ns 3 C++ simulation code.

I want it to include a dataset of prompts and reference solutions, grouped by difficulty, with several networking topics like wifi, tcp, udp, and similar. It should support different prompt styles, like direct instruction, step by step reasoning, few examples, and self review. I also want scripts that can generate code from a model, save the outputs, and then evaluate them with both similarity metrics and runnable test cases.

Please make it easy to run the benchmark end to end, from loading prompts to scoring results, and include clear documentation for how to install dependencies and run the evaluation. If you need to check current docs or library details online, go ahead and do that.