Diffusion Beats Autoregressive in Data-Constrained Settings
arxiv.org · 7,915 words · saved by 1 readers
N/A
Diffusion Beats Autoregressive in Data-Constrained Settings Mihir Prabhudesai∗ Mengning Wu∗ Amir Zadeh Carnegie Mellon University Carnegie Mellon University Lambda Katerina Fragkiadaki Deepak Pathak arXiv:2507.15857v7 [cs.LG] 26 Oct 2025…
related reading
- Large Language Diffusion Modelsarxiv.org
- Yudhister Kumaryudhister.me
- 2406.07524arxiv.org
- What are Diffusion Models? | Lil'Loglilianweng.github.io
- 2503.09573arxiv.org
- [2510.03280] Training Optimal Large Diffusion Language Modelsar5iv.labs.arxiv.org
- Kuleshov Group | How to Build a Diffusion Language Modelkuleshov-group.github.io
- Diffusion is spectral autoregression – Sander Dielemansander.ai
- What are Diffusion Models?lilianweng.github.io
- Diffusion language models – Sander Dielemansander.ai
- 2409.02908arxiv.org
- Scaling Laws, Carefully | Lil'Loglilianweng.github.io