ImportError: cannot import name 'rank_zero_only' from 'pytorch_lightning.utilities.distributed'

[rexroth0619]

I've been running WebUI no problem in a docker container, within a 3.10.6 conda environment, but this error suddenly appeared. I did use another dreambooth ipynb and did some training -- did it alter my environment somehow? Has anyone encountered this problem before?

Also after install pytorch-lightining again in conda environment, it returns a new error saying
ImportError: cannot import name 'Mapping' from 'collections' (/opt/conda/envs/webui/lib/python3.10/collections/init.py)

I checked online and some says pytorch-lightning doesn't work with 3.10, but problem is A1111's wiki says everything is built upon 3.10.6, and I've been running it for a month without a problem. Very confused here

---

ImportError Traceback (most recent call last)
File /workspace/stable-diffusion-webui/launch.py:249
247 if name == "main":
248 prepare_enviroment()
--> 249 start_webui()

File /workspace/stable-diffusion-webui/launch.py:243, in start_webui()
241 def start_webui():
242 print(f"Launching Web UI with arguments: {' '.join(sys.argv[1:])}")
--> 243 import webui
244 webui.webui()

File /workspace/stable-diffusion-webui/webui.py:12
8 from fastapi.middleware.gzip import GZipMiddleware
10 from modules.paths import script_path
---> 12 from modules import devices, sd_samplers, upscaler, extensions
13 import modules.codeformer_model as codeformer
14 import modules.extras

File /workspace/stable-diffusion-webui/modules/sd_samplers.py:11
9 import ldm.models.diffusion.ddim
10 import ldm.models.diffusion.plms
---> 11 from modules import prompt_parser, devices, processing, images
13 from modules.shared import opts, cmd_opts, state
14 import modules.shared as shared

File /workspace/stable-diffusion-webui/modules/processing.py:14
11 from skimage import exposure
12 from typing import Any, Dict, List, Optional
---> 14 import modules.sd_hijack
15 from modules import devices, prompt_parser, masking, sd_samplers, lowvram, generation_parameters_copypaste
16 from modules.sd_hijack import model_hijack

File /workspace/stable-diffusion-webui/modules/sd_hijack.py:10
7 from torch import einsum
8 from torch.nn.functional import silu
---> 10 import modules.textual_inversion.textual_inversion
11 from modules import prompt_parser, devices, sd_hijack_optimizations, shared
12 from modules.shared import opts, device, cmd_opts

File /workspace/stable-diffusion-webui/modules/textual_inversion/textual_inversion.py:13
9 import csv
11 from PIL import Image, PngImagePlugin
---> 13 from modules import shared, devices, sd_hijack, processing, sd_models, images
14 import modules.textual_inversion.dataset
15 from modules.textual_inversion.learn_schedule import LearnRateScheduler

File /workspace/stable-diffusion-webui/modules/shared.py:14
12 import modules.interrogate
13 import modules.memmon
---> 14 import modules.sd_models
15 import modules.styles
16 import modules.devices as devices

File /workspace/stable-diffusion-webui/modules/sd_models.py:14
12 from modules import shared, modelloader, devices, script_callbacks
13 from modules.paths import models_path
---> 14 from modules.sd_hijack_inpainting import do_inpainting_hijack, should_hijack_inpainting
16 model_dir = "Stable-diffusion"
17 model_path = os.path.abspath(os.path.join(models_path, model_dir))

File /workspace/stable-diffusion-webui/modules/sd_hijack_inpainting.py:6
3 from einops import repeat
4 from omegaconf import ListConfig
----> 6 import ldm.models.diffusion.ddpm
7 import ldm.models.diffusion.ddim
8 import ldm.models.diffusion.plms

File /workspace/stable-diffusion-webui/repositories/stable-diffusion/ldm/models/diffusion/ddpm.py:19
17 from tqdm import tqdm
18 from torchvision.utils import make_grid
---> 19 from pytorch_lightning.utilities.distributed import rank_zero_only
21 from ldm.util import log_txt_as_img, exists, default, ismap, isimage, mean_flat, count_params, instantiate_from_config
22 from ldm.modules.ema import LitEma

ImportError: cannot import name 'rank_zero_only' from 'pytorch_lightning.utilities.distributed' (/opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/utilities/distributed.py)

---

---

ImportError Traceback (most recent call last)
File /workspace/stable-diffusion-webui/launch.py:249
247 if name == "main":
248 prepare_enviroment()
--> 249 start_webui()

File /workspace/stable-diffusion-webui/launch.py:243, in start_webui()
241 def start_webui():
242 print(f"Launching Web UI with arguments: {' '.join(sys.argv[1:])}")
--> 243 import webui
244 webui.webui()

File /workspace/stable-diffusion-webui/webui.py:12
8 from fastapi.middleware.gzip import GZipMiddleware
10 from modules.paths import script_path
---> 12 from modules import devices, sd_samplers, upscaler, extensions
13 import modules.codeformer_model as codeformer
14 import modules.extras

File /workspace/stable-diffusion-webui/modules/sd_samplers.py:11
9 import ldm.models.diffusion.ddim
10 import ldm.models.diffusion.plms
---> 11 from modules import prompt_parser, devices, processing, images
13 from modules.shared import opts, cmd_opts, state
14 import modules.shared as shared

File /workspace/stable-diffusion-webui/modules/processing.py:14
11 from skimage import exposure
12 from typing import Any, Dict, List, Optional
---> 14 import modules.sd_hijack
15 from modules import devices, prompt_parser, masking, sd_samplers, lowvram, generation_parameters_copypaste
16 from modules.sd_hijack import model_hijack

File /workspace/stable-diffusion-webui/modules/sd_hijack.py:10
7 from torch import einsum
8 from torch.nn.functional import silu
---> 10 import modules.textual_inversion.textual_inversion
11 from modules import prompt_parser, devices, sd_hijack_optimizations, shared
12 from modules.shared import opts, device, cmd_opts

File /workspace/stable-diffusion-webui/modules/textual_inversion/textual_inversion.py:13
9 import csv
11 from PIL import Image, PngImagePlugin
---> 13 from modules import shared, devices, sd_hijack, processing, sd_models, images
14 import modules.textual_inversion.dataset
15 from modules.textual_inversion.learn_schedule import LearnRateScheduler

File /workspace/stable-diffusion-webui/modules/shared.py:14
12 import modules.interrogate
13 import modules.memmon
---> 14 import modules.sd_models
15 import modules.styles
16 import modules.devices as devices

File /workspace/stable-diffusion-webui/modules/sd_models.py:14
12 from modules import shared, modelloader, devices, script_callbacks
13 from modules.paths import models_path
---> 14 from modules.sd_hijack_inpainting import do_inpainting_hijack, should_hijack_inpainting
16 model_dir = "Stable-diffusion"
17 model_path = os.path.abspath(os.path.join(models_path, model_dir))

File /workspace/stable-diffusion-webui/modules/sd_hijack_inpainting.py:6
3 from einops import repeat
4 from omegaconf import ListConfig
----> 6 import ldm.models.diffusion.ddpm
7 import ldm.models.diffusion.ddim
8 import ldm.models.diffusion.plms

File /workspace/stable-diffusion-webui/repositories/stable-diffusion/ldm/models/diffusion/ddpm.py:12
10 import torch.nn as nn
11 import numpy as np
---> 12 import pytorch_lightning as pl
13 from torch.optim.lr_scheduler import LambdaLR
14 from einops import rearrange, repeat

File /opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/init.py:53
50 sys.stdout.write(f'Partial import of {__name__} during the build process.\n') # pragma: no-cover
51 # We are not importing the rest of the lightning during the build process, as it may not be compiled yet
52 else:
---> 53 from pytorch_lightning.core import LightningModule, data_loader
54 from pytorch_lightning.callbacks import Callback
55 from pytorch_lightning.trainer import Trainer

File /opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/core/init.py:339
1 """
2 A :class:~LightningModule organizes your PyTorch code into the following sections:
3
(...)
336
337 """
--> 339 from pytorch_lightning.core.decorators import data_loader
340 from pytorch_lightning.core.lightning import LightningModule
342 all = ['LightningModule', 'data_loader']

File /opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/core/decorators.py:6
2 from typing import Callable
4 import torch
----> 6 from pytorch_lightning.core.lightning import LightningModule
7 from pytorch_lightning.utilities import rank_zero_warn
10 def data_loader(fn):

File /opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/core/lightning.py:19
17 from pytorch_lightning import _logger as log
18 from pytorch_lightning.core.grads import GradInformation
---> 19 from pytorch_lightning.core.hooks import ModelHooks
20 from pytorch_lightning.core.memory import ModelSummary
21 from pytorch_lightning.core.saving import ModelIO, PRIMITIVE_TYPES, ALLOWED_CONFIG_TYPES

File /opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/core/hooks.py:7
5 from torch.nn import Module
6 from torch.optim.optimizer import Optimizer
----> 7 from pytorch_lightning.utilities import move_data_to_device, NATIVE_AMP_AVALAIBLE
10 try:
11 from apex import amp

File /opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/utilities/init.py:7
4 import torch
6 from pytorch_lightning.utilities.distributed import rank_zero_only, rank_zero_warn, rank_zero_info
----> 7 from pytorch_lightning.utilities.apply_func import move_data_to_device
8 from pytorch_lightning.utilities.parsing import AttributeDict
10 try:

File /opt/conda/envs/webui/lib/python3.10/site-packages/pytorch_lightning/utilities/apply_func.py:2
1 from abc import ABC
----> 2 from collections import Mapping, Sequence
3 from copy import copy
4 from typing import Any, Callable, Union

ImportError: cannot import name 'Mapping' from 'collections' (/opt/conda/envs/webui/lib/python3.10/collections/init.py)
​

[pbonazzi]

The function was moved to a new script with a different name . Try to import it as :

from pytorch_lightning.utilities.rank_zero import rank_zero_only

[thekahal]

Hi there.
I get the same error by using an online notebook created by someone else and that is open to the public.
https://colab.research.google.com/drive/1lx9AGsrh7MlyJhK9UrNTK8pYpARnx457?usp=sharing
Can the issue be fixed here? If so, how? Can I do it - or only the author can do it? If I can do it - can you please let me know how?
Thank you.

[visoutre]

@thekahal
you can fix this yourself using pbonazzi's solution above. It's only 1 line of code and in the case of your colab, the line causing trouble is line 12 in this file

But manually fixing the file each time you load the colab is not an efficient solution, just a work around.

I don't think CompVis is going to fix the line on their end. Another solution may be to install an older version of Pytorch Lightning (before they changed this code). I'll try to contact someone working on Pytorch Lightning and see if they have a better solution that doesn't involve editing the CompVis code.

[limitsnova]

I edited the main.py and removed line 12 / added line 13

still errors received

from pytorch_lightning.utilities.distributed import rank_zero_only
ImportError: cannot import name 'rank_zero_only' from 'pytorch_lightning.utilities.distributed' (/home/simon/miniconda3/envs/automatic/lib/python3.9/site-packages/pytorch_lightning/utilities/distributed.py)

any workaround ? I am using local , not collab.
Thank you

[thekahal]

@visoutre Thank you so much - your solution worked! THANK YOU!!!

[limitsnova]

Fixed my issue using Conda install today.