Abra: Scaling Diffusion Image Training
- Published
- Source
- arXiv
- Paper number
- 945
- Field
- Machine Learning
- arXiv ID
- 2608.17286
Key points
- A controlled model family spanning 10^19 to 10^22 FLOPs was used to measure the compute-optimal point for text-to-image diffusion models.
- The optimum is about 200 image tokens per parameter, ten times the Chinchilla rule for language models at 20 tokens per parameter.
- Diffusion models are robust to overtraining, so when compute remains available, using more data is preferable to increasing model size.
- Generative-quality metrics such as FID and CLIPScore, representation quality, and optimal CFG settings also vary predictably with compute.
- The compute-optimal token count rises with resolution, making high-resolution training compute-bound rather than data-bound.
Paper links
External research summaries. These are not HDATF publications or measured product results.