# Stable Diffusion v1.5 converted to LiteRT

This repository contains a LiteRT/TFLite export of the Hugging Face model `stable-diffusion-v1-5/stable-diffusion-v1-5`.

## Variants

- `fp32/`: reference export
- `int8/`: mixed bundle with fp32 text encoder fallback, PT2E dynamic int8 UNet, and fp32 VAE fallback

## Files per variant

- `text_encoder.tflite`
- `unet.tflite`
- `vae_decoder.tflite`

## Shared assets

- `tokenizer/`
- `scheduler/`
- `configs/`
- `conversion_manifest.json`

## Notes

- Stable Diffusion v1.5 is a multi-stage pipeline, so this export is split into submodels.
- Both the `fp32` and `int8` variants are smoke-tested by reloading the serialized LiteRT models and executing inference.
- The preview images in `preview/` are decoder smoke tests, not final text-to-image samples.