In this repository I present a pretrained 33.7 million parameter RNA model, based on the novel xLSTM architecture. The model is pretrained on RNAcentral database, and later evaluated on tasks of RNA secondary structure prediction, splice site prediction and mean ribosome loading. The work was done for my master thesis at Faculty of Electrical Engineering and Computing and internship at Genome Institute of Singapore.

Biorxiv paper: RNA-xLSTM: Evaluating xLSTM as an Alternative Foundation to Transformers in RNA Modeling

NOTE: Pretraining script and data code is not available for public use, this repository contains all other parts of the project.