Jmlr_giantstep

Our paper How two-layer neural networks learn, one (giant) step at a time was accepted for publication at JMLR! It will also feature as a poster at ICLR in the JMLR track.