stable-diffusion-webui-docker ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports'

Has this issue been opened before?

[x] It is not in the FAQ, I checked.
[x] It is not in the issues, I searched.

Describe the bug

I think this is just me. But running into an issue and hoping someone else knows how to fix it. Can't find the issue online. Anyway. Stable diffusion web-ui has been working perfectly on my machine for ages, but recently tried to start it again and am getting stuck with this error:

webui-docker-invoke-1  | Traceback (most recent call last):
webui-docker-invoke-1  |   File "/opt/conda/bin/invokeai-configure", line 5, in <module>
webui-docker-invoke-1  |     from ldm.invoke.config.invokeai_configure import main
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/config/invokeai_configure.py", line 40, in <module>
webui-docker-invoke-1  |     from ..args import PRECISION_CHOICES, Args
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/args.py", line 100, in <module>
webui-docker-invoke-1  |     from ldm.invoke.conditioning import split_weighted_subprompts
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/conditioning.py", line 18, in <module>
webui-docker-invoke-1  |     from .generator.diffusers_pipeline import StableDiffusionGeneratorPipeline
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/generator/__init__.py", line 4, in <module>
webui-docker-invoke-1  |     from .base import Generator
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/generator/base.py", line 21, in <module>
webui-docker-invoke-1  |     from pytorch_lightning import seed_everything
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/__init__.py", line 34, in <module>
webui-docker-invoke-1  |     from pytorch_lightning.callbacks import Callback  # noqa: E402
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/__init__.py", line 25, in <module>
webui-docker-invoke-1  |     from pytorch_lightning.callbacks.progress import ProgressBarBase, RichProgressBar, TQDMProgressBar
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/progress/__init__.py", line 22, in <module>
webui-docker-invoke-1  |     from pytorch_lightning.callbacks.progress.rich_progress import RichProgressBar  # noqa: F401
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/progress/rich_progress.py", line 20, in <module>
webui-docker-invoke-1  |     from torchmetrics.utilities.imports import _compare_version
webui-docker-invoke-1  | ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports' (/opt/conda/lib/python3.10/site-packages/torchmetrics/utilities/imports.py)

Which UI

invoke

Hardware / Software

OS: Linux Mint
OS version: 21.1
WSL version (if applicable):
Docker Version: 24.0.6
Docker compose version: v2.21.0
Repo version: master 6a34739135eb112667f00943c1fac98ab294716a
RAM: 48GB
GPU/VRAM: nvidia 2060

Steps to Reproduce

docker compose --profile download up --build docker compose --profile invoke up --build

Additional context

I already made a fresh clone and cleaned my docker containers (docker system prune -a). So this is a completely fresh build.

Anyone run into this? Any pointers?

Over at invoke ai I found something about installing torchmetrics v0.11.4 (https://github.com/invoke-ai/InvokeAI/issues/3658). Is this something I can configure with an env var?

Oct 04 '23 08:10 dinandmentink

Yup just faced the exact same issue with the Invoke profile. Changed to auto profile for now - which works.

Oct 07 '23 12:10 jaredquekjz

me,too.

Oct 19 '23 04:10 DoiiarX

Same here. Steps to reproduce:

git clone https://github.com/AbdBarho/stable-diffusion-webui-docker.git
cd stable-diffusion-webui-docker/
docker compose --profile download up --build
docker compose --profile invoke up --build

Using a Amazon g5.xlarge instance (NVIDIA A10G Tensor-Core-GPU) with an EC2 Deep Learning Base GPU AMI (Ubuntu 20.04) 20231026 (ami-0d134e01570c1e7b4)

$ docker --version
Docker version 24.0.6, build ed223bc
$ uname -a
Linux ip-xxx-xxx-xxx-xxx 5.15.0-1048-aws #53~20.04.1-Ubuntu SMP Wed Oct 4 16:44:20 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux

I've also tried git checkout tags/8.1.0, but got the same error

:~/stable-diffusion-webui-docker$ docker compose --profile invoke up --build
[+] Building 0.7s (17/17) FINISHED                                                                                                            docker:default
 => [invoke internal] load build definition from Dockerfile                                                                                             0.0s
 => => transferring dockerfile: 1.99kB                                                                                                                  0.0s
 => [invoke internal] load .dockerignore                                                                                                                0.0s
 => => transferring context: 2B                                                                                                                         0.0s
 => [invoke internal] load metadata for docker.io/pytorch/pytorch:2.0.1-cuda11.7-cudnn8-runtime                                                         0.6s
 => [invoke internal] load metadata for docker.io/library/alpine:3.17                                                                                   0.7s
 => [invoke stage-1 1/8] FROM docker.io/pytorch/pytorch:2.0.1-cuda11.7-cudnn8-runtime@sha256:82e0d379a5dedd6303c89eda57bcc434c40be11f249ddfadfd5673b84  0.0s
 => [invoke internal] load build context                                                                                                                0.0s
 => => transferring context: 65B                                                                                                                        0.0s
 => [invoke xformers 1/3] FROM docker.io/library/alpine:3.17@sha256:f71a5f071694a785e064f05fed657bf8277f1b2113a8ed70c90ad486d6ee54dc                    0.0s
 => CACHED [invoke stage-1 2/8] RUN --mount=type=cache,target=/var/cache/apt   apt-get update &&   apt-get install make g++ git libopencv-dev -y &&     0.0s
 => CACHED [invoke stage-1 3/8] RUN git clone https://github.com/invoke-ai/InvokeAI.git /InvokeAI                                                       0.0s
 => CACHED [invoke stage-1 4/8] WORKDIR /InvokeAI                                                                                                       0.0s
 => CACHED [invoke stage-1 5/8] RUN --mount=type=cache,target=/root/.cache/pip   git reset --hard f3b2e02921927d9317255b1c3811f47bd40a2bf9 &&   pip in  0.0s
 => CACHED [invoke stage-1 6/8] RUN --mount=type=cache,target=/root/.cache/pip   git fetch &&   git reset --hard &&   git checkout main &&   git reset  0.0s
 => CACHED [invoke xformers 2/3] RUN apk add --no-cache aria2                                                                                           0.0s
 => CACHED [invoke xformers 3/3] RUN aria2c -x 5 --dir / --out wheel.whl 'https://github.com/AbdBarho/stable-diffusion-webui-docker/releases/download/  0.0s
 => CACHED [invoke stage-1 7/8] RUN --mount=type=cache,target=/root/.cache/pip   --mount=type=bind,from=xformers,source=/wheel.whl,target=/xformers-0.  0.0s
 => CACHED [invoke stage-1 8/8] COPY . /docker/                                                                                                         0.0s
 => [invoke] exporting to image                                                                                                                         0.0s
 => => exporting layers                                                                                                                                 0.0s
 => => writing image sha256:effc0d511e7589ea6981692f8685c58396379348bbc89cd8adac14bb4191848d                                                            0.0s
 => => naming to docker.io/library/sd-invoke:30                                                                                                         0.0s
[+] Running 1/0
 ✔ Container webui-docker-invoke-1  Created                                                                                                             0.0s
Attaching to webui-docker-invoke-1
webui-docker-invoke-1  | Mounted ldm
webui-docker-invoke-1  | Mounted .cache
webui-docker-invoke-1  | Mounted RealESRGAN
webui-docker-invoke-1  | Mounted Codeformer
webui-docker-invoke-1  | Mounted GFPGAN
webui-docker-invoke-1  | Mounted GFPGANv1.4.pth
webui-docker-invoke-1  | Loading Python libraries...
webui-docker-invoke-1  |
webui-docker-invoke-1  | Traceback (most recent call last):
webui-docker-invoke-1  |   File "/opt/conda/bin/invokeai-configure", line 5, in <module>
webui-docker-invoke-1  |     from ldm.invoke.config.invokeai_configure import main
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/config/invokeai_configure.py", line 40, in <module>
webui-docker-invoke-1  |     from ..args import PRECISION_CHOICES, Args
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/args.py", line 100, in <module>
webui-docker-invoke-1  |     from ldm.invoke.conditioning import split_weighted_subprompts
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/conditioning.py", line 18, in <module>
webui-docker-invoke-1  |     from .generator.diffusers_pipeline import StableDiffusionGeneratorPipeline
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/generator/__init__.py", line 4, in <module>
webui-docker-invoke-1  |     from .base import Generator
webui-docker-invoke-1  |   File "/InvokeAI/ldm/invoke/generator/base.py", line 21, in <module>
webui-docker-invoke-1  |     from pytorch_lightning import seed_everything
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/__init__.py", line 34, in <module>
webui-docker-invoke-1  |     from pytorch_lightning.callbacks import Callback  # noqa: E402
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/__init__.py", line 25, in <module>
webui-docker-invoke-1  |     from pytorch_lightning.callbacks.progress import ProgressBarBase, RichProgressBar, TQDMProgressBar
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/progress/__init__.py", line 22, in <module>
webui-docker-invoke-1  |     from pytorch_lightning.callbacks.progress.rich_progress import RichProgressBar  # noqa: F401
webui-docker-invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/progress/rich_progress.py", line 20, in <module>
webui-docker-invoke-1  |     from torchmetrics.utilities.imports import _compare_version
webui-docker-invoke-1  | ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports' (/opt/conda/lib/python3.10/site-packages/torchmetrics/utilities/imports.py)
webui-docker-invoke-1 exited with code 1
:~/stable-diffusion-webui-docker$

Nov 04 '23 15:11 cyril23

#596

Nov 13 '23 20:11 AbdBarho

Hmm, if I understand this correctly, this error should not be happening anymore, right?

Well, I just ran into it with a fresh install today.

Dec 25 '23 20:12 gitwittidbit

Same here with invoke

Dec 30 '23 21:12 ap1969

Ditto here on a local install following instructions. And the exact same procedure described above to reproduce it

Jan 11 '24 06:01 aravindprasad

Anyone still getting this error - you can modify the dockerfile at \stable-diffusion-webui-docker\services\invoke\Dockerfile by adding a line: WORKDIR ${ROOT}

RUN pip install torchmetrics==0.11.4

RUN --mount=type=cache,target=/root/.cache/pip
git reset --hard f3b2e02921927d9317255b1c3811f47bd40a2bf9 &&
pip install -e .

Jan 14 '24 12:01 whereismyfun42

RUN pip install torchmetrics==0.11.4

Brilliant - thanks!

But now I'm facing another error: cannot import name 'ModelSearchArguments' from 'huggingface_hub'

Jan 29 '24 21:01 gitwittidbit

RUN pip install torchmetrics==0.11.4

same here, did you solve it?

Feb 04 '24 13:02 TIMESTICKING

Yup just faced the exact same issue with the Invoke profile. Changed to auto profile for now - which works.

Hi, I wonder if the already downloaded files will be removed automaticlly whenI switch "invoke" to "auto", because it is kinda waste of space since I would never use "invoke" again. Or how should I removed the files that related to the "invoke" mode. THX

Feb 04 '24 13:02 TIMESTICKING

Describe the bug

[+] Running 1/1
 ✔ Container webui-docker-invoke-1  Created                                                                                                                                                 0.1s
Attaching to invoke-1
invoke-1  | mkdir: created directory '/data/.cache/invoke'
invoke-1  | mkdir: created directory '/data/.cache/invoke/ldm/'
invoke-1  | Mounted ldm
invoke-1  | Mounted .cache
invoke-1  | Mounted RealESRGAN
invoke-1  | mkdir: created directory '/data/models/Codeformer/'
invoke-1  | Mounted Codeformer
invoke-1  | Mounted GFPGAN
invoke-1  | Mounted GFPGANv1.4.pth
invoke-1  | Loading Python libraries...
invoke-1  |
invoke-1  | Traceback (most recent call last):
invoke-1  |   File "/opt/conda/bin/invokeai-configure", line 5, in <module>
invoke-1  |     from ldm.invoke.config.invokeai_configure import main
invoke-1  |   File "/InvokeAI/ldm/invoke/config/invokeai_configure.py", line 40, in <module>
invoke-1  |     from ..args import PRECISION_CHOICES, Args
invoke-1  |   File "/InvokeAI/ldm/invoke/args.py", line 100, in <module>
invoke-1  |     from ldm.invoke.conditioning import split_weighted_subprompts
invoke-1  |   File "/InvokeAI/ldm/invoke/conditioning.py", line 18, in <module>
invoke-1  |     from .generator.diffusers_pipeline import StableDiffusionGeneratorPipeline
invoke-1  |   File "/InvokeAI/ldm/invoke/generator/__init__.py", line 4, in <module>
invoke-1  |     from .base import Generator
invoke-1  |   File "/InvokeAI/ldm/invoke/generator/base.py", line 21, in <module>
invoke-1  |     from pytorch_lightning import seed_everything
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/__init__.py", line 34, in <module>
invoke-1  |     from pytorch_lightning.callbacks import Callback  # noqa: E402
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/__init__.py", line 25, in <module>
invoke-1  |     from pytorch_lightning.callbacks.progress import ProgressBarBase, RichProgressBar, TQDMProgressBar
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/progress/__init__.py", line 22, in <module>
invoke-1  |     from pytorch_lightning.callbacks.progress.rich_progress import RichProgressBar  # noqa: F401
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/pytorch_lightning/callbacks/progress/rich_progress.py", line 20, in <module>
invoke-1  |     from torchmetrics.utilities.imports import _compare_version
invoke-1  | ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports' (/opt/conda/lib/python3.10/site-packages/torchmetrics/utilities/imports.py)
invoke-1 exited with code 1

Which UI

invoke

Hardware / Software

OS: Windows 11 Pro
OS version: Version 23H2 (OS Build 22631.3296)
WSL version NAME STATE VERSION
- docker-desktop Running 2 docker-desktop-data Running 2
Docker Version: Client: Cloud integration: v1.0.35+desktop.11 Version: 25.0.3 API version: 1.44 Go version: go1.21.6 Git commit: 4debf41 Built: Tue Feb 6 21:13:02 2024 OS/Arch: windows/amd64 Context: default Server: Docker Desktop 4.28.0 (139021) Engine: Version: 25.0.3 API version: 1.44 (minimum version 1.24) Go version: go1.21.6 Git commit: f417435 Built: Tue Feb 6 21:14:25 2024 OS/Arch: linux/amd64 Experimental: false containerd: Version: 1.6.28 GitCommit: ae07eda36dd25f8a1b98dfbf587313b99c0190bb runc: Version: 1.1.12 GitCommit: v1.1.12-0-g51d5e94 docker-init: Version: 0.19.0 GitCommit: de40ad0
Docker compose version: Docker Compose version v2.24.6-desktop.1
Repo version: 1.2.0
RAM: 64GB
GPU/VRAM: RTX 4090/24GB

Steps to Reproduce

run setup commands"

docker compose --profile download up --build
docker compose --profile invoke up --build

Error:

invoke-1  | ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports' (/opt/conda/lib/python3.10/site-packages/torchmetrics/utilities/imports.py)
invoke-1 exited with code 1

Container runs for 47 seconds and stops

Additional context new install with

git clone https://github.com/AbdBarho/stable-diffusion-webui-docker.git

Mar 18 '24 15:03 kkthebeast

RUN pip install torchmetrics==0.11.4

Brilliant - thanks!

But now I'm facing another error: cannot import name 'ModelSearchArguments' from 'huggingface_hub'

huggingface_hub removed 'ModelSearchArguments' from version v0.19, you can remove the current version and re-install v0.18.0 or below, please note that you also need to lower the version of transformers to v4.35. Otherwise you will get "AttributeError: module 'huggingface_hub.constants' has no attribute 'HF_HUB_CACHE'" error again.

I solved this issue by adding :

RUN --mount=type=cache,target=/root/.cache/pip \
  pip uninstall -y torchmetrics && \
  pip install torchmetrics==0.11.4 && \
  pip uninstall -y huggingface-hub && \
  pip install huggingface-hub==0.18.0 && \
  pip uninstall -y transformers && \
  pip install transformers==4.35.2

Mar 30 '24 15:03 mynameiskeen

Welp @mynameiskeen , your changes seems to almost solve the problem in my case 😅 now I got this:

invoke-1  | * --web was specified, starting web server...
invoke-1  | Traceback (most recent call last):
invoke-1  |   File "/opt/conda/bin/invokeai", line 8, in <module>
invoke-1  |     sys.exit(main())
invoke-1  |   File "/InvokeAI/ldm/invoke/CLI.py", line 184, in main
invoke-1  |     invoke_ai_web_server_loop(gen, gfpgan, codeformer, esrgan)
invoke-1  |   File "/InvokeAI/ldm/invoke/CLI.py", line 1078, in invoke_ai_web_server_loop
invoke-1  |     from invokeai.backend import InvokeAIWebServer
invoke-1  |   File "/InvokeAI/invokeai/backend/__init__.py", line 4, in <module>
invoke-1  |     from .invoke_ai_web_server import InvokeAIWebServer
invoke-1  |   File "/InvokeAI/invokeai/backend/invoke_ai_web_server.py", line 17, in <module>
invoke-1  |     from flask import Flask, redirect, send_from_directory, request, make_response
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/__init__.py", line 7, in <module>
invoke-1  |     from .app import Flask as Flask
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/app.py", line 27, in <module>
invoke-1  |     from . import cli
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/cli.py", line 17, in <module>
invoke-1  |     from .helpers import get_debug_flag
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/helpers.py", line 14, in <module>
invoke-1  |     from werkzeug.urls import url_quote
invoke-1  | ImportError: cannot import name 'url_quote' from 'werkzeug.urls' (/opt/conda/lib/python3.10/site-packages/werkzeug/urls.py)
invoke-1  | Exception ignored in atexit callback: <built-in function write_history_file>
invoke-1  | FileNotFoundError: [Errno 2] No such file or directory
invoke-1 exited with code 1

Apr 01 '24 12:04 dakyskye

Welp @mynameiskeen , your changes seems to almost solve the problem in my case 😅 now I got this:

invoke-1  | * --web was specified, starting web server...
invoke-1  | Traceback (most recent call last):
invoke-1  |   File "/opt/conda/bin/invokeai", line 8, in <module>
invoke-1  |     sys.exit(main())
invoke-1  |   File "/InvokeAI/ldm/invoke/CLI.py", line 184, in main
invoke-1  |     invoke_ai_web_server_loop(gen, gfpgan, codeformer, esrgan)
invoke-1  |   File "/InvokeAI/ldm/invoke/CLI.py", line 1078, in invoke_ai_web_server_loop
invoke-1  |     from invokeai.backend import InvokeAIWebServer
invoke-1  |   File "/InvokeAI/invokeai/backend/__init__.py", line 4, in <module>
invoke-1  |     from .invoke_ai_web_server import InvokeAIWebServer
invoke-1  |   File "/InvokeAI/invokeai/backend/invoke_ai_web_server.py", line 17, in <module>
invoke-1  |     from flask import Flask, redirect, send_from_directory, request, make_response
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/__init__.py", line 7, in <module>
invoke-1  |     from .app import Flask as Flask
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/app.py", line 27, in <module>
invoke-1  |     from . import cli
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/cli.py", line 17, in <module>
invoke-1  |     from .helpers import get_debug_flag
invoke-1  |   File "/opt/conda/lib/python3.10/site-packages/flask/helpers.py", line 14, in <module>
invoke-1  |     from werkzeug.urls import url_quote
invoke-1  | ImportError: cannot import name 'url_quote' from 'werkzeug.urls' (/opt/conda/lib/python3.10/site-packages/werkzeug/urls.py)
invoke-1  | Exception ignored in atexit callback: <built-in function write_history_file>
invoke-1  | FileNotFoundError: [Errno 2] No such file or directory
invoke-1 exited with code 1

Try add this: pip uninstall -y Werkzeug &&
pip install Werkzeug==2.2.2

Apr 19 '24 13:04 mynameiskeen

stable-diffusion-webui-docker stable-diffusion-webui-docker copied to clipboard

ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports'

stable-diffusion-webui-docker
stable-diffusion-webui-docker copied to clipboard