Official code for ReLoRA from the paper Stack More Layers Differently: High-Rank Training Through Low-Rank Updates - View it on GitHub