Spaces:
Running
Running
amazonaws-la
commited on
Commit
•
5b7982f
1
Parent(s):
8343f48
Update app.py
Browse files
app.py
CHANGED
@@ -10,7 +10,7 @@ import numpy as np
|
|
10 |
import PIL.Image
|
11 |
import spaces
|
12 |
import torch
|
13 |
-
from diffusers import AutoencoderKL, DiffusionPipeline
|
14 |
|
15 |
DESCRIPTION = "# SDXL"
|
16 |
if not torch.cuda.is_available():
|
@@ -23,6 +23,7 @@ USE_TORCH_COMPILE = os.getenv("USE_TORCH_COMPILE") == "1"
|
|
23 |
ENABLE_CPU_OFFLOAD = os.getenv("ENABLE_CPU_OFFLOAD") == "1"
|
24 |
ENABLE_REFINER = os.getenv("ENABLE_REFINER", "1") == "1"
|
25 |
ENABLE_USE_LORA = os.getenv("ENABLE_USE_LORA", "1") == "1"
|
|
|
26 |
|
27 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
28 |
|
@@ -48,15 +49,25 @@ def generate(
|
|
48 |
guidance_scale_refiner: float = 5.0,
|
49 |
num_inference_steps_base: int = 25,
|
50 |
num_inference_steps_refiner: int = 25,
|
|
|
51 |
use_lora: bool = False,
|
52 |
apply_refiner: bool = False,
|
53 |
-
model = '
|
54 |
-
vaecall = '
|
55 |
lora = 'amazonaws-la/juliette',
|
|
|
56 |
) -> PIL.Image.Image:
|
57 |
if torch.cuda.is_available():
|
58 |
-
pipe = DiffusionPipeline.from_pretrained(model, torch_dtype=torch.float16)
|
59 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
if use_lora:
|
61 |
pipe.load_lora_weights(lora)
|
62 |
pipe.fuse_lora(lora_scale=0.7)
|
@@ -134,6 +145,13 @@ with gr.Blocks(css="style.css") as demo:
|
|
134 |
model = gr.Text(label='Modelo')
|
135 |
vaecall = gr.Text(label='VAE')
|
136 |
lora = gr.Text(label='LoRA')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
137 |
with gr.Row():
|
138 |
prompt = gr.Text(
|
139 |
label="Prompt",
|
@@ -191,6 +209,7 @@ with gr.Blocks(css="style.css") as demo:
|
|
191 |
step=32,
|
192 |
value=1024,
|
193 |
)
|
|
|
194 |
use_lora = gr.Checkbox(label='Use Lora', value=False, visible=ENABLE_USE_LORA)
|
195 |
apply_refiner = gr.Checkbox(label="Apply refiner", value=False, visible=ENABLE_REFINER)
|
196 |
with gr.Row():
|
@@ -253,6 +272,13 @@ with gr.Blocks(css="style.css") as demo:
|
|
253 |
queue=False,
|
254 |
api_name=False,
|
255 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
256 |
use_lora.change(
|
257 |
fn=lambda x: gr.update(visible=x),
|
258 |
inputs=use_lora,
|
@@ -298,11 +324,13 @@ with gr.Blocks(css="style.css") as demo:
|
|
298 |
guidance_scale_refiner,
|
299 |
num_inference_steps_base,
|
300 |
num_inference_steps_refiner,
|
|
|
301 |
use_lora,
|
302 |
apply_refiner,
|
303 |
model,
|
304 |
vaecall,
|
305 |
lora,
|
|
|
306 |
],
|
307 |
outputs=result,
|
308 |
api_name="run",
|
|
|
10 |
import PIL.Image
|
11 |
import spaces
|
12 |
import torch
|
13 |
+
from diffusers import DPMSolverMultistepScheduler, AutoencoderKL, DiffusionPipeline
|
14 |
|
15 |
DESCRIPTION = "# SDXL"
|
16 |
if not torch.cuda.is_available():
|
|
|
23 |
ENABLE_CPU_OFFLOAD = os.getenv("ENABLE_CPU_OFFLOAD") == "1"
|
24 |
ENABLE_REFINER = os.getenv("ENABLE_REFINER", "1") == "1"
|
25 |
ENABLE_USE_LORA = os.getenv("ENABLE_USE_LORA", "1") == "1"
|
26 |
+
ENABLE_USE_VAE = os.getenv("ENABLE_USE_VAE", "1") == "1"
|
27 |
|
28 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
29 |
|
|
|
49 |
guidance_scale_refiner: float = 5.0,
|
50 |
num_inference_steps_base: int = 25,
|
51 |
num_inference_steps_refiner: int = 25,
|
52 |
+
use_vae: bool = False,
|
53 |
use_lora: bool = False,
|
54 |
apply_refiner: bool = False,
|
55 |
+
model = 'SG161222/Realistic_Vision_V6.0_B1_noVAE',
|
56 |
+
vaecall = 'stabilityai/sd-vae-ft-mse',
|
57 |
lora = 'amazonaws-la/juliette',
|
58 |
+
lora_scale: float = 0.7,
|
59 |
) -> PIL.Image.Image:
|
60 |
if torch.cuda.is_available():
|
|
|
61 |
|
62 |
+
if not use_vae:
|
63 |
+
pipe = DiffusionPipeline.from_pretrained(model, torch_dtype=torch.float16)
|
64 |
+
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
|
65 |
+
|
66 |
+
if use_vae:
|
67 |
+
vae = AutoencoderKL.from_pretrained(vaecall, torch_dtype=torch.float16)
|
68 |
+
pipe = DiffusionPipeline.from_pretrained(model, vae=vae, torch_dtype=torch.float16)
|
69 |
+
pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
|
70 |
+
|
71 |
if use_lora:
|
72 |
pipe.load_lora_weights(lora)
|
73 |
pipe.fuse_lora(lora_scale=0.7)
|
|
|
145 |
model = gr.Text(label='Modelo')
|
146 |
vaecall = gr.Text(label='VAE')
|
147 |
lora = gr.Text(label='LoRA')
|
148 |
+
lora_scale = gr.Slider(
|
149 |
+
label="Lora Scale",
|
150 |
+
minimum=0.01,
|
151 |
+
maximum=1,
|
152 |
+
step=0.01,
|
153 |
+
value=0.7,
|
154 |
+
)
|
155 |
with gr.Row():
|
156 |
prompt = gr.Text(
|
157 |
label="Prompt",
|
|
|
209 |
step=32,
|
210 |
value=1024,
|
211 |
)
|
212 |
+
use_vae = gr.Checkbox(label='Use VAE', value=False, visible=ENABLE_USE_VAE)
|
213 |
use_lora = gr.Checkbox(label='Use Lora', value=False, visible=ENABLE_USE_LORA)
|
214 |
apply_refiner = gr.Checkbox(label="Apply refiner", value=False, visible=ENABLE_REFINER)
|
215 |
with gr.Row():
|
|
|
272 |
queue=False,
|
273 |
api_name=False,
|
274 |
)
|
275 |
+
use_vae.change(
|
276 |
+
fn=lambda x: gr.update(visible=x),
|
277 |
+
inputs=use_vae,
|
278 |
+
outputs=vaecall,
|
279 |
+
queue=False,
|
280 |
+
api_name=False,
|
281 |
+
)
|
282 |
use_lora.change(
|
283 |
fn=lambda x: gr.update(visible=x),
|
284 |
inputs=use_lora,
|
|
|
324 |
guidance_scale_refiner,
|
325 |
num_inference_steps_base,
|
326 |
num_inference_steps_refiner,
|
327 |
+
use_vae,
|
328 |
use_lora,
|
329 |
apply_refiner,
|
330 |
model,
|
331 |
vaecall,
|
332 |
lora,
|
333 |
+
lora_scale,
|
334 |
],
|
335 |
outputs=result,
|
336 |
api_name="run",
|