File size: 1,573 Bytes
179eb90 bc9726e 179eb90 bc9726e 179eb90 bc9726e 179eb90 ca10f59 179eb90 ca10f59 179eb90 ca10f59 179eb90 ca10f59 179eb90 ca10f59 179eb90 bc9726e 179eb90 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 |
---
tags:
- text-to-image
- stable-diffusion
- lora
- diffusers
base_model: stabilityai/stable-diffusion-2-1-base
license: mit
library_name: diffusers
---
# Model description
Official SD21(base) Model of the paper [Trajectory Consistency Distillation](https://arxiv.org/abs/2402.19159).
For more usage please found at [Project Page](https://huggingface.co/h1t/TCD-SDXL-LoRA/)
Here is a simple example:
```python
import torch
from diffusers import StableDiffusionPipeline, TCDScheduler
device = "cuda"
base_model_id = "stabilityai/stable-diffusion-2-1-base"
tcd_lora_id = "h1t/TCD-SD21-base-LoRA"
pipe = StableDiffusionPipeline.from_pretrained(base_model_id, torch_dtype=torch.float16, variant="fp16").to(device)
pipe.scheduler = TCDScheduler.from_config(pipe.scheduler.config)
pipe.load_lora_weights(tcd_lora_id)
pipe.fuse_lora()
prompt = "Beautiful woman, bubblegum pink, lemon yellow, minty blue, futuristic, high-detail, epic composition, watercolor."
image = pipe(
prompt=prompt,
num_inference_steps=4,
guidance_scale=0,
# Eta (referred to as `gamma` in the paper) is used to control the stochasticity in every step.
# A value of 0.3 often yields good results.
# We recommend using a higher eta when increasing the number of inference steps.
eta=0.3,
generator=torch.Generator(device=device).manual_seed(0),
).images[0]
```
|