The model is the GPT_neo with 1,2B parameters. Was fine-tuned at 10% sample from SO dataset for QA task.
Base Promt: "Question: ...\nAnswer:"
