**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 15, 2025, 5:53am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F5 "2025-05-15T05:53:56Z")
still getting a plain black image, but no errors
```auto
import logging
from diffusers import AutoPipelineForText2Image, AutoencoderKL
import torch
import numpy as np
import random
import os
from PIL import Image
# =========================
# STEP 0: Logging Setup
# =========================
log_file = "generation_log.txt"
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[
logging.FileHandler(log_file),
logging.StreamHandler()
]
)
logging.info("Initializing...")
# =========================
# STEP 1: Environment Setup
# =========================
torch.cuda.empty_cache()
torch.cuda.ipc_collect()
seed = random.randint(0, 9999999)
torch.manual_seed(seed)
np.random.seed(seed)
logging.info(f"Using seed: {seed}")
# ===============================
# STEP 2: Model and LoRA Setup
# ===============================
logging.info("Loading base model and LoRA weights...")
model_dir = "D:\\Ganu\\AIImage\\huggingface\\kohya_ss\\kohya_ss\\outputs"
lora_weights_path = os.path.join(model_dir, "model")
model_id = "stabilityai/stable-diffusion-xl-base-1.0"
# Optional: Custom VAE (uncomment if needed)
vae = AutoencoderKL.from_pretrained(
"madebyollin/sdxl-vae-fp16-fix",
torch_dtype=torch.float16
).to("cuda")
try:
pipeline = AutoPipelineForText2Image.from_pretrained(
model_id,
torch_dtype=torch.float16,
variant="fp16",
#device_map="auto"
).to("cuda")
logging.info("Pipeline loaded to GPU with float16.")
except Exception as e:
logging.error(f"Failed to load model pipeline: {e}")
raise
# pipeline.enable_model_cpu_offload()
# If using VAE:
pipeline.vae = vae
pipeline.enable_attention_slicing()
pipeline.enable_vae_slicing()
try:
pipeline.load_lora_weights(lora_weights_path, weight_name="last.safetensors")
logging.info("LoRA weights loaded successfully.")
except ValueError as e:
logging.error("Invalid LoRA checkpoint. Check the format or compatibility.")
raise e
# =========================
# STEP 3: Prompt Inference
# =========================
text_prompt = (
"A wide, breathtaking landscape with all real vibrant nature-themed background, lush forests, mountains, and a Doctor standing prominently in the foreground"
)
negative_prompt = (
"text, letters, words, signage, logos, labels, writing, messy background, busy layout, clutter, double faces, abstract shapes, UI panels with words, overlapping elements, header, footer, top bar, navigation bar, bottom menu, toolbar, top text, website layout, browser frame, button row, page border, UI bar"
)
logging.info(f"Running inference with prompt: {text_prompt}")
try:
result = pipeline(
prompt=text_prompt,
negative_prompt=negative_prompt,
guidance_scale=7.5,
num_inference_steps=30
)
generated_image = result.images[0]
output_path = f"generated_image_{seed}.png"
generated_image.save(output_path)
logging.info(f"Image saved to: {output_path}")
generated_image.show()
except Exception as e:
logging.error(f"Error during image generation: {e}")
raise
```
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 15, 2025, 6:15am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F6 "2025-05-15T06:15:00Z")
Hmm… Perhaps LoRA loading issue…?
```py
import logging
from diffusers import AutoPipelineForText2Image, AutoencoderKL
import torch
import numpy as np
import random
import os
from PIL import Image
# =========================
# STEP 0: Logging Setup
# =========================
log_file = "generation_log.txt"
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[
logging.FileHandler(log_file),
logging.StreamHandler()
]
)
logging.info("Initializing...")
# =========================
# STEP 1: Environment Setup
# =========================
torch.cuda.empty_cache()
torch.cuda.ipc_collect()
seed = random.randint(0, 9999999)
torch.manual_seed(seed)
np.random.seed(seed)
logging.info(f"Using seed: {seed}")
# ===============================
# STEP 2: Model and LoRA Setup
# ===============================
logging.info("Loading base model and LoRA weights...")
model_dir = "D:\\Ganu\\AIImage\\huggingface\\kohya_ss\\kohya_ss\\outputs"
lora_weights_path = os.path.join(model_dir, "model")
model_id = "stabilityai/stable-diffusion-xl-base-1.0"
# Optional: Custom VAE (uncomment if needed)
vae = AutoencoderKL.from_pretrained(
"madebyollin/sdxl-vae-fp16-fix",
torch_dtype=torch.float16
).to("cuda")
try:
pipeline = AutoPipelineForText2Image.from_pretrained(
model_id,
torch_dtype=torch.float16,
variant="fp16",
).to("cuda")
logging.info("Pipeline loaded to GPU with float16.")
except Exception as e:
logging.error(f"Failed to load model pipeline: {e}")
raise
#pipeline.enable_model_cpu_offload()
# If using VAE:
pipeline.vae = vae
pipeline.enable_attention_slicing()
pipeline.enable_vae_slicing()
"""try:
pipeline.load_lora_weights(lora_weights_path, weight_name="last.safetensors")
logging.info("LoRA weights loaded successfully.")
except ValueError as e:
logging.error("Invalid LoRA checkpoint. Check the format or compatibility.")
raise e
"""
# =========================
# STEP 3: Prompt Inference
# =========================
text_prompt = (
"A wide, breathtaking landscape with all real vibrant nature-themed background, lush forests, mountains, and a Doctor standing prominently in the foreground"
)
negative_prompt = (
"text, letters, words, signage, logos, labels, writing, messy background, busy layout, clutter, double faces, abstract shapes, UI panels with words, overlapping elements, header, footer, top bar, navigation bar, bottom menu, toolbar, top text, website layout, browser frame, button row, page border, UI bar"
)
logging.info(f"Running inference with prompt: {text_prompt}")
try:
result = pipeline(
prompt=text_prompt,
negative_prompt=negative_prompt,
guidance_scale=7.5,
num_inference_steps=30
)
generated_image = result.images[0]
output_path = f"generated_image_{seed}.png"
generated_image.save(output_path)
logging.info(f"Image saved to: {output_path}")
generated_image.show()
except Exception as e:
logging.error(f"Error during image generation: {e}")
raise
```
---
**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 15, 2025, 9:06am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F7 "2025-05-15T09:06:59Z")
Hello
Getting same black image without LORA
Python & Torch Details:
```auto
(venv) D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo>python
Python 3.10.10 (tags/v3.10.10:aad5f6a, Feb 7 2023, 17:20:36) [MSC v.1929 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>>
>>> # Check if torch was built with CUDA
>>> print("Is CUDA available? :", torch.cuda.is_available())
Is CUDA available? : True
>>> print("CUDA version (torch compiled with):", torch.version.cuda)
CUDA version (torch compiled with): 11.8
>>> print("Torch built with CUDA support:", torch.backends.cuda.is_built())
Torch built with CUDA support: True
```
Code:
```auto
import logging
from diffusers import AutoPipelineForText2Image, AutoencoderKL
import torch
import numpy as np
import random
import os
from PIL import Image
# =========================
# STEP 0: Logging Setup
# =========================
log_file = "generation_log.txt"
logging.basicConfig(
level=logging.INFO,
format='%(asctime)s - %(levelname)s - %(message)s',
handlers=[
logging.FileHandler(log_file),
logging.StreamHandler()
]
)
logging.info("Initializing...")
# =========================
# STEP 1: Environment Setup
# =========================
torch.cuda.empty_cache()
torch.cuda.ipc_collect()
seed = random.randint(0, 9999999)
torch.manual_seed(seed)
np.random.seed(seed)
logging.info(f"Using seed: {seed}")
# ===============================
# STEP 2: Model and LoRA Setup
# ===============================
logging.info("Loading base model and LoRA weights...")
model_dir = "D:\\Ganu\\AIImage\\huggingface\\kohya_ss\\kohya_ss\\outputs"
lora_weights_path = os.path.join(model_dir, "model")
model_id = "stabilityai/stable-diffusion-xl-base-1.0"
# Optional: Custom VAE (uncomment if needed)
vae = AutoencoderKL.from_pretrained(
"madebyollin/sdxl-vae-fp16-fix",
torch_dtype=torch.float16
).to("cuda")
try:
pipeline = AutoPipelineForText2Image.from_pretrained(
model_id,
torch_dtype=torch.float16,
variant="fp16",
).to("cuda")
logging.info("Pipeline loaded to GPU with float16.")
except Exception as e:
logging.error(f"Failed to load model pipeline: {e}")
raise
#pipeline.enable_model_cpu_offload()
# If using VAE:
pipeline.vae = vae
pipeline.enable_attention_slicing()
pipeline.enable_vae_slicing()
"""try:
pipeline.load_lora_weights(lora_weights_path, weight_name="last.safetensors")
logging.info("LoRA weights loaded successfully.")
except ValueError as e:
logging.error("Invalid LoRA checkpoint. Check the format or compatibility.")
raise e
"""
# =========================
# STEP 3: Prompt Inference
# =========================
text_prompt = (
"A wide, breathtaking landscape with all real vibrant nature-themed background, lush forests, mountains, and a Doctor standing prominently in the foreground"
)
negative_prompt = (
"text, letters, words, signage, logos, labels, writing, messy background, busy layout, clutter, double faces, abstract shapes, UI panels with words, overlapping elements, header, footer, top bar, navigation bar, bottom menu, toolbar, top text, website layout, browser frame, button row, page border, UI bar"
)
logging.info(f"Running inference with prompt: {text_prompt}")
try:
result = pipeline(
prompt=text_prompt,
negative_prompt=negative_prompt,
guidance_scale=7.5,
num_inference_steps=30
)
generated_image = result.images[0]
output_path = f"generated_image_{seed}.png"
generated_image.save(output_path)
logging.info(f"Image saved to: {output_path}")
generated_image.show()
except Exception as e:
logging.error(f"Error during image generation: {e}")
raise
```
Environment:
**1. Libraries**
```auto
pip list
Package Version
------------------ ------------------
accelerate 0.21.0
aiofiles 24.1.0
annotated-types 0.7.0
anyio 4.9.0
certifi 2025.1.31
charset-normalizer 3.4.1
click 8.1.8
colorama 0.4.6
deepspeed 0.10.0+f5c834a6
diffusers 0.21.4
exceptiongroup 1.2.2
fastapi 0.115.12
ffmpy 0.5.0
filelock 3.18.0
flash-attention 1.0.0
fsspec 2025.3.2
gradio 5.27.1
gradio_client 1.9.1
groovy 0.1.2
h11 0.16.0
hjson 3.1.0
httpcore 1.0.9
httpx 0.28.1
huggingface-hub 0.16.4
idna 3.10
importlib_metadata 8.6.1
Jinja2 3.1.6
markdown-it-py 3.0.0
MarkupSafe 3.0.2
mdurl 0.1.2
mpmath 1.3.0
mypy_extensions 1.1.0
networkx 3.4.2
ninja 1.11.1.4
numpy 1.23.1
orjson 3.10.16
packaging 25.0
pandas 2.2.3
peft 0.15.2
pillow 11.2.1
pip 25.1.1
psutil 7.0.0
py-cpuinfo 9.0.0
pydantic 1.10.13
pydantic_core 2.33.1
pydub 0.25.1
Pygments 2.19.1
pyre-extensions 0.0.29
python-dateutil 2.9.0.post0
python-multipart 0.0.20
pytz 2025.2
PyYAML 6.0.2
regex 2024.11.6
requests 2.32.3
rich 14.0.0
ruff 0.11.7
safehttpx 0.1.6
safetensors 0.5.3
semantic-version 2.10.0
setuptools 65.5.0
shellingham 1.5.4
six 1.17.0
sniffio 1.3.1
starlette 0.46.2
sympy 1.14.0
tokenizers 0.13.3
tomlkit 0.13.2
torch 2.4.0+cu118
torchaudio 2.7.0+cu118
torchvision 0.22.0+cu118
tqdm 4.67.1
transformers 4.31.0
typer 0.15.3
typing_extensions 4.13.2
typing-inspect 0.9.0
typing-inspection 0.4.0
tzdata 2025.2
urllib3 2.4.0
uvicorn 0.34.2
websockets 15.0.1
xformers 0.0.27.post2+cu118
zipp 3.21.0
```
1. nvdia-sim output
```auto
(venv) D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo>nvidia-smi
Thu May 15 14:36:50 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 576.40 Driver Version: 576.40 CUDA Version: 12.9 |
|-----------------------------------------+------------------------+----------------------+
| GPU Name Driver-Model | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+========================+======================|
| 0 NVIDIA GeForce GTX 1650 WDDM | 00000000:01:00.0 On | N/A |
| N/A 47C P8 5W / 50W | 698MiB / 4096MiB | 3% Default |
| | | N/A |
+-----------------------------------------+------------------------+----------------------+
+-----------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=========================================================================================|
| 0 N/A N/A 5648 C+G ...IA app\CEF\NVIDIA Overlay.exe N/A |
| 0 N/A N/A 6664 C+G ...we\Microsoft.Media.Player.exe N/A |
| 0 N/A N/A 7348 C+G ...Chrome\Application\chrome.exe N/A |
| 0 N/A N/A 7560 C+G ...Chrome\Application\chrome.exe N/A |
| 0 N/A N/A 7940 C+G ....0.3240.64\msedgewebview2.exe N/A |
| 0 N/A N/A 9104 C+G C:\Windows\explorer.exe N/A |
| 0 N/A N/A 9520 C+G ...h_cw5n1h2txyewy\SearchApp.exe N/A |
| 0 N/A N/A 9728 C+G ...ntrolPanel\SystemSettings.exe N/A |
| 0 N/A N/A 11968 C+G ...h_cw5n1h2txyewy\SearchApp.exe N/A |
| 0 N/A N/A 14064 C+G ...5n1h2txyewy\TextInputHost.exe N/A |
| 0 N/A N/A 15292 C+G ...IA app\CEF\NVIDIA Overlay.exe N/A |
+-----------------------------------------------------------------------------------------+
```
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 15, 2025, 9:12am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F8 "2025-05-15T09:12:51Z")
Hmm… Or CUDA Toolkit version issue?
```auto
accelerate 1.0.1
diffusers 0.32.2
torch 2.4.0+cu124
transformers 4.49.0.dev0
```
---
**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 15, 2025, 9:29am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F9 "2025-05-15T09:29:23Z")
The simple code
```auto
from diffusers import DiffusionPipeline
import torch
pipe = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0",
torch_dtype=torch.float32
).to("cuda")
prompt = "A clear sunny landscape with mountains and a river"
image = pipe(prompt=prompt).images[0]
image.save("test_image.png")
```
returns the error
```auto
(venv) D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo>python John-Training-15thMay2025.py
[2025-05-15 14:53:54,560] [INFO] [real_accelerator.py:133:get_accelerator] Setting ds_accelerator to cuda (auto detect)
D:\Ganu\AIImage\venv\lib\site-packages\deepspeed-0.10.0+f5c834a6-py3.10.egg\deepspeed\runtime\zero\linear.py:49: FutureWarning: `torch.cuda.amp.custom_fwd(args...)` is deprecated. Please use `torch.amp.custom_fwd(args..., device_type='cuda')` instead.
def forward(ctx, input, weight, bias=None):
D:\Ganu\AIImage\venv\lib\site-packages\deepspeed-0.10.0+f5c834a6-py3.10.egg\deepspeed\runtime\zero\linear.py:67: FutureWarning: `torch.cuda.amp.custom_bwd(args...)` is deprecated. Please use `torch.amp.custom_bwd(args..., device_type='cuda')` instead.
def backward(ctx, grad_output):
Loading pipeline components...: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:02<00:00, 2.48it/s]
Traceback (most recent call last):
File "D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo\John-Training-15thMay2025.py", line 7, in
).to("cuda")
File "D:\Ganu\AIImage\venv\lib\site-packages\diffusers\pipelines\pipeline_utils.py", line 733, in to
module.to(torch_device, torch_dtype)
File "D:\Ganu\AIImage\venv\lib\site-packages\transformers\modeling_utils.py", line 1900, in to
return super().to(*args, **kwargs)
File "D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py", line 1174, in to
return self._apply(convert)
File "D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py", line 780, in _apply
module._apply(fn)
File "D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py", line 780, in _apply
module._apply(fn)
File "D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py", line 780, in _apply
module._apply(fn)
File "D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py", line 805, in _apply
param_applied = fn(param)
File "D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py", line 1160, in convert
return t.to(
torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 242.00 MiB. GPU 0 has a total capacity of 4.00 GiB of which 0 bytes is free. Of the allocated memory 10.41 GiB is allocated by PyTorch, and 262.02 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
```
and chatgpt says
```auto
You're running out of VRAM on your GPU (4 GB total, 0 bytes free), and you're trying to load Stable Diffusion XL (SDXL), which requires at least 12 GB of GPU memory, ideally more. The error is expected — SDXL is far too heavy for a 4 GB GPU.
```
P.s:
1. The cpu version still works.
2. And I remember the gpu version working earlier
3. Do you want me to change the ( CUDA Toolkit version )
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 15, 2025, 10:17am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F10 "2025-05-15T10:17:53Z")
Upgrading the CUDA toolkit is a last resort, so I’ll try these first. Also, the above error may simply be due to insufficient VRAM…
```auto
pip install -U accelerate
pip install diffusers==0.32.2 transformers<=4.48.3
```
---
**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 15, 2025, 10:40am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F11 "2025-05-15T10:40:11Z")
pip install diffusers==0.32.2 transformers\<=4.48.3
The system cannot find the file specified.
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 15, 2025, 10:43am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F12 "2025-05-15T10:43:15Z")
Hmm…
```auto
pip install diffusers==0.32.2
pip install transformers==4.48.3
```
---
**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 15, 2025, 11:15am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F13 "2025-05-15T11:15:46Z")
(venv) D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo\>python John-Training-15thMay2025.py
WARNING[XFORMERS]: xFormers can’t load C++/CUDA extensions. xFormers was built for:
PyTorch 2.4.0+cu118 with CUDA 1108 (you have 2.7.0+cu118)
Python 3.10.11 (you have 3.10.10)
Please reinstall xformers (see [GitHub - facebookresearch/xformers: Hackable and optimized Transformers building blocks, supporting a composable construction.](https://github.com/facebookresearch/xformers#installing-xformers))
Memory-efficient attention, SwiGLU, sparse and more won’t be available.
Set XFORMERS\_MORE\_DETAILS=1 for more details
Loading pipeline components…: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:02\<00:00, 3.32it/s]
Traceback (most recent call last):
File “D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo\John-Training-15thMay2025.py”, line 7, in
).to(“cuda”)
File “D:\Ganu\AIImage\venv\lib\site-packages\diffusers\pipelines\pipeline\_utils.py”, line 461, in to
module.to(device, dtype)
File “D:\Ganu\AIImage\venv\lib\site-packages\transformers\modeling\_utils.py”, line 3110, in to
return super().to(\*args, \*\*kwargs)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 1355, in to
return self.\_apply(convert)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 915, in \_apply
module.\_apply(fn)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 915, in \_apply
module.\_apply(fn)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 915, in \_apply
module.\_apply(fn)
[Previous line repeated 3 more times]
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 942, in \_apply
param\_applied = fn(param)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 1341, in convert
return t.to(
torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 26.00 MiB. GPU 0 has a total capacity of 4.00 GiB of which 0 bytes is free. Of the allocated memory 10.66 GiB is allocated by PyTorch, and 239.73 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH\_CUDA\_ALLOC\_CONF=expandable\_segments:True to avoid fragmentation. See documentation for Memory Management ([CUDA semantics — PyTorch 2.7 documentation](https://pytorch.org/docs/stable/notes/cuda.html#environment-variables))
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 15, 2025, 11:19am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F14 "2025-05-15T11:19:01Z")
> WARNING[XFORMERS]: xFormers can’t load C++/CUDA extensions. xFormers was built for:
Oh…
```auto
pip uninstall xformers
```
---
**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 15, 2025, 11:21am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F15 "2025-05-15T11:21:31Z")
(venv) D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo\>python John-Training-15thMay2025.py
Loading pipeline components…: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:02\<00:00, 3.35it/s]
Traceback (most recent call last):
File “D:\Ganu\AIImage\project\Train-10Images-chatgptParameters\runs\1sstrun-23thApril2025\generation\1stGo\John-Training-15thMay2025.py”, line 7, in
).to(“cuda”)
File “D:\Ganu\AIImage\venv\lib\site-packages\diffusers\pipelines\pipeline\_utils.py”, line 461, in to
module.to(device, dtype)
File “D:\Ganu\AIImage\venv\lib\site-packages\diffusers\models\modeling\_utils.py”, line 1077, in to
return super().to(\*args, \*\*kwargs)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 1355, in to
return self.\_apply(convert)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 915, in \_apply
module.\_apply(fn)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 915, in \_apply
module.\_apply(fn)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 915, in \_apply
module.\_apply(fn)
[Previous line repeated 2 more times]
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 942, in \_apply
param\_applied = fn(param)
File “D:\Ganu\AIImage\venv\lib\site-packages\torch\nn\modules\module.py”, line 1341, in convert
return t.to(
torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 MiB. GPU 0 has a total capacity of 4.00 GiB of which 0 bytes is free. Of the allocated memory 10.67 GiB is allocated by PyTorch, and 229.42 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH\_CUDA\_ALLOC\_CONF=expandable\_segments:True to avoid fragmentation. See documentation for Memory Management ([CUDA semantics — PyTorch 2.7 documentation](https://pytorch.org/docs/stable/notes/cuda.html#environment-variables))
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 15, 2025, 11:41am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F16 "2025-05-15T11:41:18Z")
Hmm…
```py
from diffusers import DiffusionPipeline
import torch
pipe = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0",
#torch_dtype=torch.float32 # too large
torch_dtype=torch.float16
).to("cuda")
prompt = "A clear sunny landscape with mountains and a river"
image = pipe(prompt=prompt).images[0]
image.save("test_image.png")
```
---
**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 16, 2025, 5:08am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F17 "2025-05-16T05:08:30Z")
Still getting a plain black image with the following code
```auto
from diffusers import DiffusionPipeline
import torch
pipe = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0",
#torch_dtype=torch.float32 # too large
torch_dtype=torch.float16
).to("cuda")
prompt = "A clear sunny landscape with mountains and a river"
image = pipe(prompt=prompt).images[0]
image.save("test_image.png")
```
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 16, 2025, 6:19am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F18 "2025-05-16T06:19:00Z")
I found an old 10x0 related issue…
```py
from diffusers import DiffusionPipeline, AutoencoderKL
import torch
torch.backends.cudnn.benchmark = True # https://github.com/huggingface/diffusers/issues/1556
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16).to("cuda")
pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", vae=vae, variant="fp16", torch_dtype=torch.float16).to("cuda")
prompt = "A clear sunny landscape with mountains and a river"
image = pipe(prompt=prompt).images[0]
image.save("test_image.png")
```
>
>
> \### Describe the bug
>
> I am new to AI in general, and I am trying to use the diff…users package by following the hugging face tutorial for running the model online and locally. But both of the methods (GPU) gives me a black image in return. However when I use the CPU mode \`pipe.to("cpu")\`, I get the results as expected.
>
> In this case I am using the
> \`CompVis/stable-diffusion-v1-4\` model that I got from using the following snippet:
>
> \> \`\`\`
> \> git lfs install
> \> git clone https://huggingface.co/CompVis/stable-diffusion-v1-4
> \> \`\`\`
>
>
>
> \### Reproduction
>
> \`\`\`
> from diffusers import StableDiffusionPipeline
> import torch
>
>
> pipe = StableDiffusionPipeline.from\_pretrained(
> "D:\\\\4\_Projects\\\\9\_AI\\\\stable-diffusion-v1-4",
> revision="fp16",
> torch\_dtype=torch.float16,
> )
>
> pipe = pipe.to("cuda")
>
> prompt = "a photo of an astronaut riding a horse on mars"
> pipe.enable\_attention\_slicing()
> image = pipe(prompt).images\[0\]
>
> image.save(f"{prompt}.png")
> \`\`\`
> \## The fixes I tried:
> 1. Editing the safety\_checker.py
>
> \> Editing the line \`has\_nsfw\_concepts = \[len(res\["bad\_concepts"\]) \> 0 for res in result\]\` to \`has\_nsfw\_concepts = \[False for res in result\]\`
>
> 2. Using a different seed (1024, 3214)
> 3. Using a different scheduler (\`LMSDiscreteScheduler\`)
>
> \### Logs
>
> \`\`\`shell
> The following is the only things I got in the terminal using the local model.
>
> 100%|██████████████████████████████████████████████████████████████████████████| 51/51 \[04:32\<00:00, 5.33s/it\]
> Potential NSFW content was detected in one or more images. A black image will be returned instead. Try again with a different prompt and/or seed.
> PS D:\\4\_Projects\\9\_AI\\text\_to\_image\_python\>
> \`\`\`
>
>
> \### System Info
>
> \- \`diffusers\` version: 0.9.0
> \- Platform: Windows-10-10.0.19044-SP0
> \- Python version: 3.10.8
> \- PyTorch version (GPU?): 1.13.0 (True)
> \- Huggingface\_hub version: 0.11.1
> \- Transformers version: 4.25.1
> \- Using GPU in script?: yes
> \- Using distributed or parallel set-up in script?: I do not understand what distributed or parallel setup could be in my script. However I've read an article where \*\*accelerate\*\* is used for distributed setup, and I do have \*\*accelerate\*\* package installed.
---
**Author:**  [@deicool](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2Fdeicool)%5C
**Post date:** [May 16, 2025, 8:55am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F19 "2025-05-16T08:55:24Z")
A green blank image is being created with the above code…
---
**Author:**  [@John6666](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FJohn6666)%5C
**Post date:** [May 16, 2025, 9:13am UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102%2F20 "2025-05-16T09:13:36Z")
Green…? It’s surreal, but it’s good to see a change. That this might be a CUDA Toolkit version issue. I’ll do some searching.
[Next page](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fprogram-not-working-on-gpu-but-works-on-cpu%2F155102.md%3Fpage%3D2)