The design of RNAs that fulfill desired functions is one of the major challenges in computational biology. The function of an RNA molecule depends on its structure and a strong structure-to-function relationship is already achieved on the secondary structure level of RNA. Therefore, computational RNA design is often interpreted as the inversion of a folding algorithm: Given a target secondary structure, find an RNA sequence that folds into the desired structure. However, existing RNA design approaches cannot invert state-of-the-art folding algorithms because they can only predict a limited set of base interactions. In this work, we propose RNAinformer, a novel generative transformer based approach to the inverse RNA folding problem. Leveraging axial attention, we are able to process secondary structures represented as adjacency matrices, which allows us to invert state-of-the-art folding algorithms. Consequently, RNAinformer is the first model capable of designing RNAs from secondary structures without base pair restrictions. We demonstrate RNAinformer’s strong performance across different RNA design benchmarks and showcase its novelty by inverting a state-of-the-art deep learning based secondary structure prediction algorithm.