https://openreview.net/forum?id=yDo1ynArjj
Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion | OpenReview
This paper presents Diffusion Forcing, a new training paradigm where a diffusion model is trained to denoise a set of tokens with independent per-token noise...
diffusion forcingfull sequencenexttokenprediction
https://openreview.net/forum?id=yDo1ynArjj&referrer=%5Bthe%20profile%20of%20Vincent%20Sitzmann%5D(%2Fprofile%3Fid%3D~Vincent_Sitzmann1)
Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion | OpenReview
This paper presents Diffusion Forcing, a new training paradigm where a diffusion model is trained to denoise a set of tokens with independent per-token noise...
diffusion forcingfull sequencenexttokenprediction
https://arxiv.org/abs/2506.08009v1
[2506.08009v1] Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion
Abstract page for arXiv paper 2506.08009v1: Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion