LoRA Land is our collection of over 25 open-source fine-tuned models that rival or outperform GPT-4.0. LoRA Land is powered by the open-source LoRAX framework and Predibase’s Serverless Fine-tuned Endpoints, showcasing how teams can serve many fine-tuned LLMs cost-effectively on a single GPU.

We’re looking to continuously expand our set of fine-tuned models on LoRA Land and would love to feature any community contributions (with credit)! If you have any highly performing, fine-tuned open source models please submit them using this form. If possible, include details on the base model, dataset, and task as well. We’ll be in touch if yours has been chosen!