@article{escapingcollapsethestrengthofweakdata, title = {Escaping Collapse: The Strength of Weak Data for Large Language Model Training}, author = {Kareem Amin and Sara Babakniya and Alex Bie and Weiwei Kong and Umar Syed and Sergei Vassilvitskii}, year = {2025}, eprint = {2502.08924}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2502.08924v1}, }