Summary

Supra2-IMG is a small text-to-image model trained from scratch for lightweight image-generation experiments. Its approximately 104M-parameter diffusion transformer uses a frozen Flan-T5 text encoder and a Stable Diffusion VAE.

Key Features

Text-to-image generation.
Approximately 104M parameters.
Diffusion-transformer architecture.
Frozen Flan-T5-Base encoder.
SD-VAE-FT-MSE image decoder.
256-by-256 output.
128-token text context.
Apache-2.0 model release.

About Supra2-IMG


Developers can inspect a compact image model and reproduce inference without deploying a multi-billion-parameter generator. Core capabilities include Text-to-image generation, Approximately 104M parameters, Diffusion-transformer architecture, and Frozen Flan-T5-Base encoder.


Public code or model weights are linked from the project, with setup instructions for running or adapting the system. Additional characteristics include SD-VAE-FT-MSE image decoder, 256-by-256 output, 128-token text context, and Apache-2.0 model release.

Get more likes & reach the top of search results by adding this button on your site!

Embed button preview - Light theme
Embed button preview - Dark theme
TurboType Banner

Subscribe to the AI Search Newsletter

Get top updates in AI to your inbox every weekend. It's free!