airoboros L2 13B 2.2.1
About
The Airoboros L2 13B 2.2.1 is a large language model (LLM) based on the Llama 2 architecture, featuring 13 billion parameters. It is primarily designed for instruction following rather than casual chat or roleplay. This model was fine-tuned using a mostly synthetic dataset generated by the airoboros tool, incorporating enhancements such as re-generated writing responses and longer contextual blocks. It is capable of performing various tasks, including question answering, summarization, and code generation. However, users should be cautious of potential hallucinations if not properly prompted. The model's efficacy heavily relies on specific prompt formats, which include delimiters for closed-context instructions. Licensing complexities arise due to the involvement of OpenAI's GPT-4 API in its training data, which might restrict commercial applications.