Spaces:
Running
on
Zero
Running
on
Zero
Update load_models_utils.py
Browse files- load_models_utils.py +61 -32
load_models_utils.py
CHANGED
@@ -1,52 +1,81 @@
|
|
1 |
import yaml
|
2 |
import torch
|
|
|
3 |
from diffusers import StableDiffusionXLPipeline
|
4 |
from utils import PhotoMakerStableDiffusionXLPipeline
|
5 |
-
import os
|
6 |
|
7 |
-
def get_models_dict():
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
try:
|
10 |
data = yaml.safe_load(stream)
|
11 |
-
|
|
|
12 |
return data
|
13 |
-
|
14 |
except yaml.YAMLError as exc:
|
15 |
-
|
16 |
|
17 |
-
def load_models(model_info,device,photomaker_path):
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
22 |
|
23 |
if model_type == "original":
|
24 |
-
|
25 |
-
pipe = StableDiffusionXLPipeline.from_single_file(
|
26 |
-
path,
|
27 |
-
torch_dtype=torch.float16
|
28 |
-
)
|
29 |
-
else:
|
30 |
-
pipe = StableDiffusionXLPipeline.from_pretrained(path, torch_dtype=torch.float16, use_safetensors=use_safetensors)
|
31 |
-
pipe = pipe.to(device)
|
32 |
elif model_type == "Photomaker":
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
pipe =
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
pipe.load_photomaker_adapter(
|
44 |
os.path.dirname(photomaker_path),
|
45 |
subfolder="",
|
46 |
weight_name=os.path.basename(photomaker_path),
|
47 |
-
trigger_word="img"
|
48 |
)
|
49 |
pipe.fuse_lora()
|
50 |
-
|
51 |
-
raise NotImplementedError("You should choice between original and Photomaker!",f"But you choice {model_type}")
|
52 |
return pipe
|
|
|
1 |
import yaml
|
2 |
import torch
|
3 |
+
import os
|
4 |
from diffusers import StableDiffusionXLPipeline
|
5 |
from utils import PhotoMakerStableDiffusionXLPipeline
|
|
|
6 |
|
7 |
+
def get_models_dict(config_path='config/models.yaml', verbose=False):
|
8 |
+
"""
|
9 |
+
Loads model configuration from a YAML file.
|
10 |
+
|
11 |
+
Args:
|
12 |
+
config_path (str): Path to the YAML configuration file.
|
13 |
+
verbose (bool): If True, prints the loaded configuration.
|
14 |
+
|
15 |
+
Returns:
|
16 |
+
dict: Parsed YAML data.
|
17 |
+
"""
|
18 |
+
if not os.path.exists(config_path):
|
19 |
+
raise FileNotFoundError(f"Config file '{config_path}' not found.")
|
20 |
+
|
21 |
+
with open(config_path, 'r') as stream:
|
22 |
try:
|
23 |
data = yaml.safe_load(stream)
|
24 |
+
if verbose:
|
25 |
+
print("Loaded model configuration:", data)
|
26 |
return data
|
|
|
27 |
except yaml.YAMLError as exc:
|
28 |
+
raise RuntimeError(f"Error parsing YAML file: {exc}")
|
29 |
|
30 |
+
def load_models(model_info, device="cuda", photomaker_path=None):
|
31 |
+
"""
|
32 |
+
Loads a Stable Diffusion XL model or a PhotoMaker variant based on the provided info.
|
33 |
+
|
34 |
+
Args:
|
35 |
+
model_info (dict): Model configuration dictionary.
|
36 |
+
device (str): Target device ('cuda' or 'cpu').
|
37 |
+
photomaker_path (str, optional): Path to PhotoMaker adapter weights if using Photomaker.
|
38 |
+
|
39 |
+
Returns:
|
40 |
+
DiffusionPipeline: Loaded diffusion pipeline.
|
41 |
+
"""
|
42 |
+
path = model_info.get("path")
|
43 |
+
single_file = model_info.get("single_files", False)
|
44 |
+
use_safetensors = model_info.get("use_safetensors", True)
|
45 |
+
model_type = model_info.get("model_type", "original")
|
46 |
+
|
47 |
+
if not path:
|
48 |
+
raise ValueError("Model path must be specified in the model_info.")
|
49 |
|
50 |
if model_type == "original":
|
51 |
+
pipeline_cls = StableDiffusionXLPipeline
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
52 |
elif model_type == "Photomaker":
|
53 |
+
pipeline_cls = PhotoMakerStableDiffusionXLPipeline
|
54 |
+
else:
|
55 |
+
raise NotImplementedError(
|
56 |
+
f"Unsupported model type '{model_type}'. Choose either 'original' or 'Photomaker'."
|
57 |
+
)
|
58 |
+
|
59 |
+
# Load model
|
60 |
+
if single_file:
|
61 |
+
print(f"Loading model from a single file: {path}")
|
62 |
+
pipe = pipeline_cls.from_single_file(path, torch_dtype=torch.float16)
|
63 |
+
else:
|
64 |
+
print(f"Loading model from a directory: {path}")
|
65 |
+
pipe = pipeline_cls.from_pretrained(path, torch_dtype=torch.float16, use_safetensors=use_safetensors)
|
66 |
+
|
67 |
+
pipe = pipe.to(device)
|
68 |
+
|
69 |
+
# Load PhotoMaker adapter if needed
|
70 |
+
if model_type == "Photomaker":
|
71 |
+
if not photomaker_path:
|
72 |
+
raise ValueError("Photomaker model type requires a valid 'photomaker_path'.")
|
73 |
pipe.load_photomaker_adapter(
|
74 |
os.path.dirname(photomaker_path),
|
75 |
subfolder="",
|
76 |
weight_name=os.path.basename(photomaker_path),
|
77 |
+
trigger_word="img"
|
78 |
)
|
79 |
pipe.fuse_lora()
|
80 |
+
|
|
|
81 |
return pipe
|