wehub-resource-sync
80 lines
1.8 KiB
Markdown
80 lines
1.8 KiB
Markdown
# Installation
|
|
|
|
## Requirements
|
|
|
|
- Python >= 3.10.0 (Recommend to use [Anaconda](https://www.anaconda.com/download/#linux) or [Miniconda](https://docs.conda.io/en/latest/miniconda.html))
|
|
- [PyTorch >= 2.5.1+cu12.4](https://pytorch.org/)
|
|
|
|
## Quick Install
|
|
|
|
```bash
|
|
git clone https://github.com/NVlabs/Sana.git
|
|
cd Sana
|
|
|
|
bash ./environment_setup.sh sana
|
|
# or you can install each components step by step following environment_setup.sh
|
|
```
|
|
|
|
## Hardware Requirements
|
|
|
|
| Model | VRAM Required |
|
|
|-------|---------------|
|
|
| Sana-0.6B | 9GB |
|
|
| Sana-1.6B | 12GB |
|
|
| 4-bit Quantized | < 8GB |
|
|
|
|
!!! Note
|
|
All the tests are done on A100 GPUs. Different GPU versions may vary.
|
|
|
|
## Diffusers Installation
|
|
|
|
To use Sana with `diffusers`, make sure to upgrade to the latest version:
|
|
|
|
```bash
|
|
pip install git+https://github.com/huggingface/diffusers
|
|
```
|
|
|
|
## Quick Start with Diffusers
|
|
|
|
```python
|
|
import torch
|
|
from diffusers import SanaPipeline
|
|
|
|
pipe = SanaPipeline.from_pretrained(
|
|
"Efficient-Large-Model/SANA1.5_1.6B_1024px_diffusers",
|
|
torch_dtype=torch.bfloat16,
|
|
)
|
|
pipe.to("cuda")
|
|
|
|
pipe.vae.to(torch.bfloat16)
|
|
pipe.text_encoder.to(torch.bfloat16)
|
|
|
|
prompt = 'a cyberpunk cat with a neon sign that says "Sana"'
|
|
image = pipe(
|
|
prompt=prompt,
|
|
height=1024,
|
|
width=1024,
|
|
guidance_scale=4.5,
|
|
num_inference_steps=20,
|
|
generator=torch.Generator(device="cuda").manual_seed(42),
|
|
)[0]
|
|
|
|
image[0].save("sana.png")
|
|
```
|
|
|
|
## Optional: Docker
|
|
|
|
```bash
|
|
# Build Docker image
|
|
docker build -t sana .
|
|
|
|
# Run inference with Docker
|
|
docker run --gpus all -it sana python scripts/inference.py
|
|
```
|
|
|
|
## Next Steps
|
|
|
|
- [Model Zoo](model_zoo.md) - Choose your model
|
|
- [SANA-Sprint](sana_sprint.md) - Fast inference mode with 1-4 steps generations
|
|
- [SANA-Video](sana_video.md) - Video Gen with Linear Attention and Linear Block KV-Cache
|