fast-stable-diffusion
fast-stable-diffusion copied to clipboard
TheLastBen
ImportError: cannot import name 'is_xformers_available' from 'diffusers.utils.import_utils'
Traceback (most recent call last):
File "/content/diffusers/examples/dreambooth/train_dreambooth.py", line 18, in
rerun the dependencies cell
I just have, plus disconnect, reconnect, start from scratch. The issue persists
The A100 GPU seems to not work with the current xformers pre-compiled files, I will fix that shortly
Awesome, thanks for the great support, I'm also looking forward this fix!
Try it now and see if it's fixed, use the official link : https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast-DreamBooth.ipynb
I just tried again with the updated notebook but still giving me the same error with a A100 on Colab. And btw thank you for the prompt support!
Yeah I checked and I indeed had an A100 instance. To double check I just tried to run the exact same notebook on a lesser Tesla T4 on Colab again and it works (although way slower of course) so it appears to be related to the A100 GPU on my side.
Yes I am getting the the exact same xformers error (with the SD model downloaded with my HF token) when starting the DreamBooth training.
oh wait, I just figured the problem Are you sure you're using my fork of diffusers ?
Try it now and see if it's fixed, use the official link : https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast-DreamBooth.ipynb
I am using the notebook from here without changing anything. When my other training jobs are finished I'll try again and send you a screenshot of the xformer cell's code if will still be of relevance. Thx again.
File "/usr/local/lib/python3.7/dist-packages/diffusers/models/attention.py", line 21, in
from diffusers.utils.import_utils import is_xformers_available
My version of the attention.py doesn't contain : from diffusers.utils.import_utils import is_xformers_available
I am sorry, but I am experiencing the same
Link is https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast-DreamBooth.ipynb#scrollTo=1-9QbkfAVYYU
Tried closing, refreshing, reconnecting...
Traceback (most recent call last):
File "/content/diffusers/examples/dreambooth/train_dreambooth.py", line 18, in
#@markdown # Dependencies %%capture %cd /content/ !git clone https://github.com/TheLastBen/diffusers !pip install -q git+https://github.com/TheLastBen/diffusers !pip install -q accelerate==0.12.0 !pip install -q OmegaConf !pip install -q wget !wget https://github.com/TheLastBen/fast-stable-diffusion/raw/main/Dreambooth/Deps !mv Deps Deps.7z !7z x Deps.7z !cp -r /content/usr/local/lib/python3.7/dist-packages /usr/local/lib/python3.7/ !rm Deps.7z !rm -r /content/usr !sed -i 's@else prefix + ": "@else prefix + ""@g' /usr/local/lib/python3.7/dist-packages/tqdm/std.py
==========
#@markdown # xformers
from subprocess import getoutput from IPython.display import HTML from IPython.display import clear_output import wget import time
s = getoutput('nvidia-smi') if 'T4' in s: gpu = 'T4' elif 'P100' in s: gpu = 'P100' elif 'V100' in s: gpu = 'V100' elif 'A100' in s: gpu = 'A100'
while True: try: gpu=='T4'or gpu=='P100'or gpu=='V100'or gpu=='A100' break except: pass print('[1;31mit seems that your GPU is not supported at the moment') time.sleep(5)
if (gpu=='T4'): %pip install -q https://github.com/TheLastBen/fast-stable-diffusion/raw/main/precompiled/T4/xformers-0.0.13.dev0-py3-none-any.whl
elif (gpu=='P100'): %pip install -q https://github.com/TheLastBen/fast-stable-diffusion/raw/main/precompiled/P100/xformers-0.0.13.dev0-py3-none-any.whl
elif (gpu=='V100'): %pip install -q https://github.com/TheLastBen/fast-stable-diffusion/raw/main/precompiled/V100/xformers-0.0.13.dev0-py3-none-any.whl
elif (gpu=='A100'): %cd /usr/local/lib/python3.7/diffusers/ !rm /usr/local/lib/python3.7/diffusers/models/attention.py wget.download('https://raw.githubusercontent.com/huggingface/diffusers/main/src/diffusers/models/attention.py')
clear_output() print('[1;32mDONE !')
File "/usr/local/lib/python3.7/dist-packages/diffusers/models/attention.py", line 21, in from diffusers.utils.import_utils import is_xformers_availableMy version of the attention.py doesn't contain :
from diffusers.utils.import_utils import is_xformers_available
it's running this attention.py, it seems like the colab doc only just broke in the past few hours?
!wget -O attention.py https://raw.githubusercontent.com/huggingface/diffusers/main/src/diffusers/models/attention.py
I'm not 100% sure as I've been running a few different things recently but pretty sure it was A100.
I think was using (Hardware Accelerator = GPU, GPU Class = Premium) with the specs below, in a colab using https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast-DreamBooth.ipynb about 6 hours ago
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.32.03 Driver Version: 460.32.03 CUDA Version: 11.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 A100-SXM4-40GB Off | 00000000:00:04.0 Off | 0 |
| N/A 30C P0 45W / 400W | 0MiB / 40536MiB | 0% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
if I wasn't using that then the settings were just (Hardware Accelerator = GPU and standard class GPU)
@TheLastBen
I Had the same problem, seems fixed now.
The training is now starting with the A100.
it seems to be fixed here too, thanks for creating this repo @TheLastBen 🙏
Can't confirm at the moment as I've started the training on the standard GPU but will report back if similar issues occur. Thank you!
Off-topic, is it just me or the standard and premium GPUs have roughly the same seconds / iteration? I don't remember the A100 training, but with standard GPU I get approx 1 sec / iteration at a cost of 1.96 computing units / hour. A100 is mind blowingly expensive at 13 units / hour, I can't afford that rate.
If I have the Pro+ subscription but use the standard GPU, do I still get background execution?
Most likely you can get background execution as it's related to your plan, not the GPU
HI,
I am getting same errror. I am using Custom GCE VM. (Created using offical colab guide.)
Traceback (most recent call last):
File "/content/diffusers/examples/dreambooth/train_dreambooth.py", line 18, in <module>
from diffusers import AutoencoderKL, DDPMScheduler, StableDiffusionPipeline, UNet2DConditionModel
File "/usr/local/lib/python3.7/dist-packages/diffusers/__init__.py", line 21, in <module>
from .models import AutoencoderKL, UNet2DConditionModel, UNet2DModel, VQModel
File "/usr/local/lib/python3.7/dist-packages/diffusers/models/__init__.py", line 19, in <module>
from .unet_2d import UNet2DModel
File "/usr/local/lib/python3.7/dist-packages/diffusers/models/unet_2d.py", line 11, in <module>
from .unet_blocks import UNetMidBlock2D, get_down_block, get_up_block
File "/usr/local/lib/python3.7/dist-packages/diffusers/models/unet_blocks.py", line 20, in <module>
from .attention import AttentionBlock, SpatialTransformer
File "/usr/local/lib/python3.7/dist-packages/diffusers/models/attention.py", line 21, in <module>
from diffusers.utils.import_utils import is_xformers_available
ImportError: cannot import name 'is_xformers_available' from 'diffusers.utils.import_utils' (/usr/local/lib/python3.7/dist-packages/diffusers/utils/import_utils.py)
Traceback (most recent call last):
File "/usr/local/bin/accelerate", line 8, in <module>
sys.exit(main())
File "/usr/local/lib/python3.7/dist-packages/accelerate/commands/accelerate_cli.py", line 43, in main
args.func(args)
File "/usr/local/lib/python3.7/dist-packages/accelerate/commands/launch.py", line 837, in launch_command
simple_launcher(args)
File "/usr/local/lib/python3.7/dist-packages/accelerate/commands/launch.py", line 354, in simple_launcher
raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
subprocess.CalledProcessError: Command '['/usr/bin/python3', '/content/diffusers/examples/dreambooth/train_dreambooth.py', '--image_captions_filename', '--train_text_encoder', '--save_starting_step=500', '--stop_text_encoder_training=1200', '--save_n_steps=0', '--pretrained_model_name_or_path=/content/stable-diffusion-v1-5', '--instance_data_dir=/content/gdrive/MyDrive/Fast-Dreambooth/Sessions/vikas/instance_images', '--output_dir=/content/models/vikas', '--instance_prompt=', '--seed=96576', '--resolution=512', '--mixed_precision=no', '--train_batch_size=1', '--gradient_accumulation_steps=1', '--use_8bit_adam', '--learning_rate=2e-6', '--lr_scheduler=polynomial', '--center_crop', '--lr_warmup_steps=0', '--max_train_steps=3000']' returned non-zero exit status 1.
Something went wrong
Wed Nov 2 21:50:54 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.32.03 Driver Version: 460.32.03 CUDA Version: 11.2 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 A100-SXM4-40GB Off | 00000000:00:04.0 Off | 0 |
| N/A 30C P0 47W / 400W | 0MiB / 40536MiB | 0% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
