1 paper · 1 filter
Hongcan Guo, Qinyu Zhao, Yian Zhao +8
Large language models have achieved remarkable success under the autoregressive paradigm, yet high-quality text generation need not be tied to a fixed left-to-right order. Existing…