No module named 'pytorch3d'

Question

No module named 'pytorch3d'

memoabuser opened this issue 2 years ago · 23 comments

memoabuser commented 2 years ago

I received those errors, and the diffuse isn't running, not really familiar with that, but I followed all the steps

Answer 1 · 2022-05-15T23:20:32.000Z

Actually, I solved most of it, But i get that error now {No module named 'pytorch3d')

Answer 2 · 2022-05-16T15:29:34.000Z

1- conda install -c fvcore -c iopath -c conda-forge fvcore iopath
2- conda install -c pytorch3d pytorch3d

solved the issue for me:
https://github.com/facebookresearch/pytorch3d/issues/1076#issuecomment-1043431615

Answer 3 · 2022-05-16T17:30:22.000Z

it gave me different issue
ModuleNotFoundError: No module named 'py3d_tools'

Answer 4 · 2022-05-16T20:18:16.000Z

solved the issue for me:
googlecolab/colabtools#2739 (comment)

Answer 5 · 2022-05-16T20:43:13.000Z

thank you so much, I solved it already and troubleshooting another one now

In cell 2. Diffusion and CLIP model settings

FileNotFoundError: [Errno 2] No such file or directory: './model/secondary_model_imagenet_2.pth'

sorry am not so familiar with coding

Answer 6 · 2022-05-16T20:46:22.000Z

and got this one too

Diffuse!

FileNotFoundError: [Errno 2] No such file or directory: './model/512x512_diffusion_uncond_finetune_008100.pt'

Answer 7 · 2022-05-16T20:49:12.000Z

you have to download them and put them in the model folder, it solved my problem:
[reddit](https://www.reddit.com/r/DiscoDiffusion/comments/uh0qg8/comment/i76awo0/?utm_source=share&utm_medium=web2x&context=3

Answer 8 · 2022-05-16T20:51:40.000Z

I will try it now, thank you

Answer 9 · 2022-05-16T22:21:37.000Z

I it worked good.. but now I have error once I run diffuse

CUDA error: unknown error
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

I tried the readme Instructions and it didn't work, or maybe I did something wrong -- I didn't find the resolution (128x128) line in settings

Answer 10 · 2022-05-16T22:22:49.000Z

NVIDIA GeForce RTX 3070
32 GB Ram
Fresh reboot

Answer 11 · 2022-05-16T22:47:08.000Z

try with a very simple prompt with just ViTB32 to true

ViTB32 = True #@param{type:"boolean"}
ViTB16 = False #@param{type:"boolean"}
ViTL14 = False #@param{type:"boolean"} # Default False
RN101 = False #@param{type:"boolean"} # Default False
RN50 = False #@param{type:"boolean"} # Default True
RN50x4 = False #@param{type:"boolean"} # Default False
RN50x16 = False #@param{type:"boolean"}
RN50x64 = False #@param{type:"boolean"}
SLIPB16 = False # param{type:"boolean"} # Default False. Looks broken, likely related to commented import of SLIP_VITB16
SLIPL16 = False # param{type:"boolean"}

Answer 12 · 2022-05-16T22:59:38.000Z

RuntimeError Traceback (most recent call last)
Input In [45], in <cell line: 153>()
151 print(diffusion_model_path)
152 model.load_state_dict(torch.load(diffusion_model_path, map_location='cpu'))
--> 153 model.requires_grad_(False).eval().to(device)
154 for name, param in model.named_parameters():
155 if 'qkv' in name or 'norm' in name or 'proj' in name:

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:899, in Module.to(self, *args, **kwargs)
895 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None,
896 non_blocking, memory_format=convert_to_format)
897 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)
--> 899 return self._apply(convert)

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:570, in Module._apply(self, fn)
568 def _apply(self, fn):
569 for module in self.children():
--> 570 module._apply(fn)
572 def compute_should_use_set_data(tensor, tensor_applied):
573 if torch._has_compatible_shallow_copy_type(tensor, tensor_applied):
574 # If the new tensor has compatible tensor type as the existing tensor,
575 # the current behavior is to change the tensor in-place using .data =,
(...)
580 # global flag to let the user control whether they want the future
581 # behavior of overwriting the existing tensor or not.

[... skipping similar frames: Module._apply at line 570 (2 times)]

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:570, in Module._apply(self, fn)
568 def _apply(self, fn):
569 for module in self.children():
--> 570 module._apply(fn)
572 def compute_should_use_set_data(tensor, tensor_applied):
573 if torch._has_compatible_shallow_copy_type(tensor, tensor_applied):
574 # If the new tensor has compatible tensor type as the existing tensor,
575 # the current behavior is to change the tensor in-place using .data =,
(...)
580 # global flag to let the user control whether they want the future
581 # behavior of overwriting the existing tensor or not.

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:593, in Module._apply(self, fn)
589 # Tensors stored in modules are graph leaves, and we don't want to
590 # track autograd history of param_applied, so we have to use
591 # with torch.no_grad():
592 with torch.no_grad():
--> 593 param_applied = fn(param)
594 should_use_set_data = compute_should_use_set_data(param, param_applied)
595 if should_use_set_data:

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:897, in Module.to..convert(t)
894 if convert_to_format is not None and t.dim() in (4, 5):
895 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None,
896 non_blocking, memory_format=convert_to_format)
--> 897 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)

RuntimeError: CUDA error: unknown error
CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect.
For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

Answer 13 · 2022-05-16T22:59:57.000Z

still giving the same error

Answer 14 · 2022-05-18T20:13:26.000Z

I didn't have that error

Answer 15 · 2022-05-19T00:19:14.000Z

yeah ok ,now I got it.
I made a New file in the note book and I copied and pasted all the cells there as the original one is read only.

Answer 16 · 2022-05-19T11:08:38.000Z

How did you solve the Permission denied???

I solved this issue by put the jupyter notebook in root folder of the repo

Answer 17 · 2022-05-19T11:12:35.000Z

RuntimeError Traceback (most recent call last) Input In [45], in <cell line: 153>() 151 print(diffusion_model_path) 152 model.load_state_dict(torch.load(diffusion_model_path, map_location='cpu')) --> 153 model.requires_grad_(False).eval().to(device) 154 for name, param in model.named_parameters(): 155 if 'qkv' in name or 'norm' in name or 'proj' in name:

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:899, in Module.to(self, *args, **kwargs) 895 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, 896 non_blocking, memory_format=convert_to_format) 897 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking) --> 899 return self._apply(convert)

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:570, in Module._apply(self, fn) 568 def _apply(self, fn): 569 for module in self.children(): --> 570 module._apply(fn) 572 def compute_should_use_set_data(tensor, tensor_applied): 573 if torch._has_compatible_shallow_copy_type(tensor, tensor_applied): 574 # If the new tensor has compatible tensor type as the existing tensor, 575 # the current behavior is to change the tensor in-place using .data =, (...) 580 # global flag to let the user control whether they want the future 581 # behavior of overwriting the existing tensor or not.

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:570, in Module._apply(self, fn) 568 def _apply(self, fn): 569 for module in self.children(): --> 570 module._apply(fn) 572 def compute_should_use_set_data(tensor, tensor_applied): 573 if torch._has_compatible_shallow_copy_type(tensor, tensor_applied): 574 # If the new tensor has compatible tensor type as the existing tensor, 575 # the current behavior is to change the tensor in-place using .data =, (...) 580 # global flag to let the user control whether they want the future 581 # behavior of overwriting the existing tensor or not.
[... skipping similar frames: Module._apply at line 570 (2 times)]
File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:570, in Module._apply(self, fn) 568 def _apply(self, fn): 569 for module in self.children(): --> 570 module._apply(fn) 572 def compute_should_use_set_data(tensor, tensor_applied): 573 if torch._has_compatible_shallow_copy_type(tensor, tensor_applied): 574 # If the new tensor has compatible tensor type as the existing tensor, 575 # the current behavior is to change the tensor in-place using .data =, (...) 580 # global flag to let the user control whether they want the future 581 # behavior of overwriting the existing tensor or not.

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:593, in Module._apply(self, fn) 589 # Tensors stored in modules are graph leaves, and we don't want to 590 # track autograd history of param_applied, so we have to use 591 # with torch.no_grad(): 592 with torch.no_grad(): --> 593 param_applied = fn(param) 594 should_use_set_data = compute_should_use_set_data(param, param_applied) 595 if should_use_set_data:

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/nn/modules/module.py:897, in Module.to..convert(t) 894 if convert_to_format is not None and t.dim() in (4, 5): 895 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, 896 non_blocking, memory_format=convert_to_format) --> 897 return t.to(device, dtype if t.is_floating_point() or t.is_complex() else None, non_blocking)

RuntimeError: CUDA error: unknown error CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

Every time I have an unknown problem restarting the ubuntu session and starting all the cells works for me.
Do you have tried with 512 * 512 models ?
Because i don't know if the 256 * 256 are downloaded too and put in model folder like 512 * 512

But for debugging ->
"Could you run your code with the CUDA_LAUNCH_BLOCKING=1 env variable and post the new stack trace please.
You can do that by running CUDA_LAUNCH_BLOCKING=1 python your_script.py."
Sauce

Ps: I run this repo with RTX3070 too without probleme

Answer 18 · 2022-05-19T11:18:09.000Z

how did you do it, Because I have the 256256 model downloaded and the 512512, I tried both and still having the same problem

Answer 19 · 2022-05-19T11:23:02.000Z

how did you do it, Because I have the 256_256 model downloaded and the 512_512, I tried both and still having the same problem

I don't know the only difference I see is that I have another version of cuda mine is 11.3 and python 3.8. But I doubt that is the problem.
What is the size of the image chosen? I am in [800, 450]

and try with prompt very simple like
[0: "a red boat"]

Answer 20 · 2022-05-19T12:49:24.000Z

Thank you, actually it worked after I changed the size into 800.450
is there no other way to increase the image size ?

Answer 21 · 2022-05-19T22:58:04.000Z

not me too, I just changed the image size in the settings cell

Answer 22 · 2022-06-17T04:03:23.000Z

solved the issue for me: googlecolab/colabtools#2739 (comment)

I've gotten as far as this, but now I get this error.
RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory

Using downloaded and verified file: ./model/superres/project.yaml/?dl=1
Using downloaded and verified file: ./model/superres/last.ckpt/?dl=1
Loading model from ./model/superres/last.ckpt/?dl=1
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
Input In [11], in <cell line: 434>()
    431     return log
    433 sr_diffMode = 'superresolution'
--> 434 sr_model = get_model('superresolution')
    436 def do_superres(img, filepath):
    438   if args.sharpen_preset == 'Faster':

Input In [11], in get_model(mode)
    213 path_conf, path_ckpt = download_models(mode)
    214 config = OmegaConf.load(path_conf)
--> 215 model, step = load_model_from_config(config, path_ckpt)
    216 return model

Input In [11], in load_model_from_config(config, ckpt)
    200 def load_model_from_config(config, ckpt):
    201     print(f"Loading model from {ckpt}")
--> 202     pl_sd = torch.load(ckpt, map_location="cpu")
    203     global_step = pl_sd["global_step"]
    204     sd = pl_sd["state_dict"]

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/serialization.py:600, in load(f, map_location, pickle_module, **pickle_load_args)
    595 if _is_zipfile(opened_file):
    596     # The zipfile reader is going to advance the current file position.
    597     # If we want to actually tail call to torch.jit.load, we need to
    598     # reset back to the original position.
    599     orig_position = opened_file.tell()
--> 600     with _open_zipfile_reader(opened_file) as opened_zipfile:
    601         if _is_torchscript_zip(opened_zipfile):
    602             warnings.warn("'torch.load' received a zip file that looks like a TorchScript archive"
    603                           " dispatching to 'torch.jit.load' (call 'torch.jit.load' directly to"
    604                           " silence this warning)", UserWarning)

File ~/anaconda3/envs/pytorch_110/lib/python3.9/site-packages/torch/serialization.py:242, in _open_zipfile_reader.__init__(self, name_or_buffer)
    241 def __init__(self, name_or_buffer) -> None:
--> 242     super(_open_zipfile_reader, self).__init__(torch._C.PyTorchFileReader(name_or_buffer))

RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory

Any idea why this would happen? Read someone who said it is due to an incomplete download, but no idea how true that is.

Answer 23 · 2023-02-08T15:38:30.000Z

Closing this as the initial issue is fixed - needed to install pytorch3d.

If you haven’t fixed this other problem feel free to open a new issue. That said - I’ve had similar issues with incomplete or corrupted models, try just deleting them and downloading again (they’re big, be patient)