GPT-Code-Clippy (GPT-CC)

GPT-Code-Clippy (GPT-CC) is an open source language model based on GPT-3. It has been fine-tuned using publicly available code from GitHub. The dataset used to train GPT-CC was created by searching SEART GitHub using specific criteria: 10+ stars, 2+ commits, a valid license and repositories that were smaller than 70708 bytes in size. This data was combined with all of the GitHub repositories found in The Pile for added accuracy. For more information about this model, please visit the following link: https://discuss.huggingface.co/t/pretrain-gpt-neo-for-open-source-githubcopilotmodel/7678?u=ncoop57