research
A Tiny Image Model Trains in 70 Minutes on Four Consumer GPUs
MiniDog pretrains a 22-million-parameter dog-image model in 55 minutes and fine-tunes it in another 15.
Summary
MiniDog pretrains a 22-million-parameter dog-image model in 55 minutes and fine-tunes it in another 15.
MiniDog is a deliberately narrow demonstration that text-to-image pretraining need not begin at frontier scale. Its authors cache image and text features, combine a small transformer with a semantic latent autoencoder and use 26,000 ImageNet dog images for pretraining plus 2,000 generated images for fine-tuning. They report 55 minutes for pretraining and 15 for fine-tuning on four RTX 3090 GPUs. The work concerns a restricted dog-breed domain and does not imply that a general-purpose generator can be trained this cheaply. Its value is pedagogical: it makes an end-to-end diffusion-transformer experiment reachable on workstation-class hardware.
Why it matters
MiniDog pretrains a 22-million-parameter dog-image model in 55 minutes and fine-tunes it in another 15.
Limits and context
- MiniDog is a deliberately narrow demonstration that text-to-image pretraining need not begin at frontier scale.
- The work concerns a restricted dog-breed domain and does not imply that a general-purpose generator can be trained this cheaply.
Key claims
MiniDog pretrains a 22-million-parameter dog-image model in 55 minutes and fine-tunes it in another 15.
Qualification: MiniDog is a deliberately narrow demonstration that text-to-image pretraining need not begin at frontier scale.
Evidence: source-2026-10-14-005
Sources
- arXiv preprint 2610.18870, original v1arXiv / submitting research authors · primary research
Corrections
No corrections have been recorded for this story.