These are the model weights associated with the TMLR 2024 publication LoRA Learns Less and Forgets Less (Biderman et al. 2024). This work was done in collaboration with Databricks Mosaic AI Research.