The goal of this project is to train and fine-tune large language models to generate benchmarks to stress-test computing systems such as compilers, operating systems and memory allocation systems.
git clone https://github.com/lac-dcc/baku.gitOn the root folder, run the command:
pip install -r requirements.txtYou can run all the scripts normal envs. But, the notebooks have very heavy Deep-Learning models, because of this, the actual experiments are setup to run in colab and generate files only on drive (but the code generation is okay for any GPU env).