File size: 2,105 Bytes
fe706b1 fb68c01 fe706b1 19d79cd fb68c01 fe706b1 fb68c01 fe706b1 19d79cd fb68c01 fe706b1 fb68c01 fe706b1 fb68c01 1828bb0 |
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 50 51 52 53 54 55 56 57 58 59 |
from typing import Dict, List, Any
import torch
from diffusers import StableDiffusionPipeline, EulerAncestralDiscreteScheduler
from PIL import Image
import base64
from io import BytesIO
# set device
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
if device.type != 'cuda':
raise ValueError("need to run on GPU")
class EndpointHandler():
def __init__(self, path=""):
# load the optimized model
self.pipe = StableDiffusionPipeline.from_pretrained(path, torch_dtype=torch.float16)
self.pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(self.pipe.scheduler.config)
self.pipe = self.pipe.to(device)
def __call__(self, data: Any) -> List[List[Dict[str, float]]]:
"""
Args:
data (:obj:):
includes the input data and the parameters for the inference.
Return:
A :obj:`dict`:. base64 encoded image
"""
prompt = data.pop("inputs", data)
params = data.pop("parameters", data)
# hyperparamters
num_inference_steps = params.pop("num_inference_steps", 20)
guidance_scale = params.pop("guidance_scale", 7.5)
negative_prompt = params.pop("negative_prompt", None)
height = params.pop("height", None)
width = params.pop("width", None)
manual_seed = params.pop("manual_seed", -1)
generator = torch.Generator(device).manual_seed(manual_seed)
# run inference pipeline
out = self.pipe(prompt,
generator=generator,
num_inference_steps=num_inference_steps,
guidance_scale=guidance_scale,
num_images_per_prompt=1,
negative_prompt=negative_prompt,
height=height,
width=width
)
# return first generate PIL image
image = out.images[0]
buffered = BytesIO()
image.save(buffered, format="JPEG")
return base64.b64encode(buffered.getvalue())
|