README.md
| 1 | |
| 2 | --- |
| 3 | license: creativeml-openrail-m |
| 4 | base_model: SG161222/Realistic_Vision_V4.0 |
| 5 | datasets: |
| 6 | - recastai/LAION-art-EN-improved-captions |
| 7 | tags: |
| 8 | - stable-diffusion |
| 9 | - stable-diffusion-diffusers |
| 10 | - text-to-image |
| 11 | - diffusers |
| 12 | inference: true |
| 13 | --- |
| 14 | |
| 15 | # Text-to-image Distillation |
| 16 | |
| 17 | This pipeline was distilled from **SG161222/Realistic_Vision_V4.0** on a Subset of **recastai/LAION-art-EN-improved-captions** dataset. Below are some example images generated with the finetuned pipeline using small-sd model. |
| 18 | |
| 19 |  |
| 20 | |
| 21 | |
| 22 | This Pipeline is based upon [the paper](https://arxiv.org/pdf/2305.15798.pdf). Training Code can be found [here](https://github.com/segmind/distill-sd). |
| 23 | |
| 24 | ## Pipeline usage |
| 25 | |
| 26 | You can use the pipeline like so: |
| 27 | |
| 28 | ```python |
| 29 | from diffusers import DiffusionPipeline |
| 30 | import torch |
| 31 | |
| 32 | pipeline = DiffusionPipeline.from_pretrained("segmind/small-sd", torch_dtype=torch.float16) |
| 33 | prompt = "Portrait of a pretty girl" |
| 34 | image = pipeline(prompt).images[0] |
| 35 | image.save("my_image.png") |
| 36 | ``` |
| 37 | |
| 38 | ## Training info |
| 39 | |
| 40 | These are the key hyperparameters used during training: |
| 41 | |
| 42 | * Steps: 95000 |
| 43 | * Learning rate: 1e-4 |
| 44 | * Batch size: 32 |
| 45 | * Gradient accumulation steps: 4 |
| 46 | * Image resolution: 512 |
| 47 | * Mixed-precision: fp16 |
| 48 | |
| 49 | |