@article{weightsubcloningdirectinitializationof, title = {Weight subcloning: direct initialization of transformers using larger pretrained ones}, author = {Mohammad Samragh and Mehrdad Farajtabar and Sachin Mehta and Raviteja Vemulapalli and Fartash Faghri and Devang Naik and Oncel Tuzel and Mohammad Rastegari}, year = {2023}, eprint = {2312.09299}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2312.09299v1}, }