Insights into a radiology-specialised multimodal large language model with sparse autoencoders

Open in new window