These are LoRA adaption weights for pretrained_model_name_or_path. The weights were fine-tuned on the fw1zr/rahul-gandhi-captions dataset.