| --- |
| title: SD1.5 Interior LoRA |
| emoji: 🚀 |
| colorFrom: indigo |
| colorTo: purple |
| sdk: gradio |
| sdk_version: 6.11.0 |
| app_file: app.py |
| pinned: false |
| license: apache-2.0 |
| tag: i |
| --- |
| |
| # LCM DreamShaper v7 — OpenVINO INT8 |
|
|
| Fast CPU-based text-to-image generation using [Latent Consistency Models](https://latent-consistency-models.github.io/) with OpenVINO INT8 quantization. |
|
|
| ## Models used |
| - [`OpenVINO/LCM_Dreamshaper_v7-int8-ov`](https://huggingface.co/OpenVINO/LCM_Dreamshaper_v7-int8-ov) — main pipeline |
| - [`rupeshs/taesd-ov`](https://huggingface.co/rupeshs/taesd-ov) — tiny autoencoder for fast decoding |
|
|
| ## Setup |
|
|
| The Space expects the model folder `LCM_Dreamshaper_v7-int8-ov` to exist at the repo root. |
| Use the download script below **once**, commit the weights, and the app will load from disk. |
|
|
| ```python |
| import huggingface_hub as hf_hub |
| |
| hf_hub.snapshot_download("OpenVINO/LCM_Dreamshaper_v7-int8-ov", local_dir="LCM_Dreamshaper_v7-int8-ov") |
| hf_hub.snapshot_download("rupeshs/taesd-ov", allow_patterns="vae_decoder/*", local_dir="LCM_Dreamshaper_v7-int8-ov") |
| hf_hub.snapshot_download("rupeshs/taesd-ov", allow_patterns="vae_encoder/*", local_dir="LCM_Dreamshaper_v7-int8-ov") |
| ``` |
|
|
| Or add a `download_models.py` script and call it from a Space startup script. |
|
|
| ## Requirements |
|
|
| ``` |
| gradio>=4.44.0 |
| openvino-genai |
| Pillow |
| numpy |
| ``` |
