Issue getting Github code to work on my own Jupyter Notebook - github

Question

I am new to Python and would like your help please.
I am copying the NorthernBobwhiteCNN code from Github to try to use the program on my computer: https://github.com/GAMELab-UGA/NorthernBobwhiteCNN. I cloned the Github files as my own Jupyter Notebook that I launched from the Command Prompt.
However, when I try to run the cells in model_prediction_example.ipynb after the import statement cell, I receive multiple errors for all the cells and the code won't run, even though everything is the exact same from Github.
Here are the errors I get using Load Trained Model cell:
RuntimeError Traceback (most recent call last)
Input In [2], in <cell line: 8>()
4 model = net.Net(params).cuda() if torch.cuda.is_available() else net.Net(params)
7 restore_path = os.path.join(model_dir, 'pretrained.pth.tar')
----> 8 _ = utils.load_checkpoint(restore_path, model, optimizer=None)
File ~\NorthernBobwhiteCNN\PythonCode\utils.py:136, in load_checkpoint(checkpoint, model, optimizer)
134 if not os.path.exists(checkpoint):
135 raise("File doesn't exist {}".format(checkpoint))
--> 136 checkpoint = torch.load(checkpoint)
137 model.load_state_dict(checkpoint['state_dict'])
139 if optimizer:
File ~\anaconda3\lib\site-packages\torch\serialization.py:789, in load(f, map_location, pickle_module, weights_only, **pickle_load_args)
787 except RuntimeError as e:
788 raise pickle.UnpicklingError(UNSAFE_MESSAGE + str(e)) from None
--> 789 return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
790 if weights_only:
791 try:
File ~\anaconda3\lib\site-packages\torch\serialization.py:1131, in _load(zip_file, map_location, pickle_module, pickle_file, **pickle_load_args)
1129 unpickler = UnpicklerWrapper(data_file, **pickle_load_args)
1130 unpickler.persistent_load = persistent_load
-> 1131 result = unpickler.load()
1133 torch._utils._validate_loaded_sparse_tensors()
1135 return result
File ~\anaconda3\lib\site-packages\torch\serialization.py:1101, in _load.<locals>.persistent_load(saved_id)
1099 if key not in loaded_storages:
1100 nbytes = numel * torch._utils._element_size(dtype)
-> 1101 load_tensor(dtype, nbytes, key, _maybe_decode_ascii(location))
1103 return loaded_storages[key]
File ~\anaconda3\lib\site-packages\torch\serialization.py:1083, in _load.<locals>.load_tensor(dtype, numel, key, location)
1079 storage = zip_file.get_storage_from_record(name, numel, torch.UntypedStorage).storage().untyped()
1080 # TODO: Once we decide to break serialization FC, we can
1081 # stop wrapping with TypedStorage
1082 loaded_storages[key] = torch.storage.TypedStorage(
-> 1083 wrap_storage=restore_location(storage, location),
1084 dtype=dtype)
File ~\anaconda3\lib\site-packages\torch\serialization.py:215, in default_restore_location(storage, location)
213 def default_restore_location(storage, location):
214 for _, _, fn in _package_registry:
--> 215 result = fn(storage, location)
216 if result is not None:
217 return result
File ~\anaconda3\lib\site-packages\torch\serialization.py:182, in _cuda_deserialize(obj, location)
180 def _cuda_deserialize(obj, location):
181 if location.startswith('cuda'):
--> 182 device = validate_cuda_device(location)
183 if getattr(obj, "_torch_load_uninitialized", False):
184 with torch.cuda.device(device):
File ~\anaconda3\lib\site-packages\torch\serialization.py:166, in validate_cuda_device(location)
163 device = torch.cuda._utils._get_device_index(location, True)
165 if not torch.cuda.is_available():
--> 166 raise RuntimeError('Attempting to deserialize object on a CUDA '
167 'device but torch.cuda.is_available() is False. '
168 'If you are running on a CPU-only machine, '
169 'please use torch.load with map_location=torch.device(\'cpu\') '
170 'to map your storages to the CPU.')
171 device_count = torch.cuda.device_count()
172 if device >= device_count:
RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.
I think that the errors are due to incorrect or missing library installations on my virtual environment.
First, I created the virtual environment "bobwhite" using conda create bobwhite in my command prompt.
Then, I did multiple conda installations based on the import statements in the model_prediction_example.ipynb.
from matplotlib import pyplot as plt
import librosa
import numpy as np
import os
from scipy import ndimage as ndi
from skimage.feature import peak_local_max
import torch
import utils
import model.net as net
So, I did the following installs in my command prompt:
conda install matplotlib
conda install -c conda-forge librosa
conda install numpy
conda install scipy
conda install scikit-image
conda install pytorch torchvision torchaudio cpuonly -c pytorch
conda install pip
pip install utils
However, I am not sure that I have installed the correct libraries to get the notebook to run. How would I find out which libraries are needed based on the import statements? Would I also need to import the libraries used in the python code net.py and utils.py as well? Additionally, I do not understand the import model.net as net statement. Is this referencing the net.py python script also found on the Github? If so, would I need to use a conda install for that, and how would I do it?

Answer 1

Change this line to
map_location = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')
checkpoint = torch.load(checkpoint, map_location=map_location)
The existing code is trying to load to cuda even if it is not available. This will make sure it loads to CPU if cuda is not available.
This part of your error's stacktrace points out the issue
File ~\NorthernBobwhiteCNN\PythonCode\utils.py:136, in load_checkpoint(checkpoint, model, optimizer)
134 if not os.path.exists(checkpoint):
135 raise("File doesn't exist {}".format(checkpoint))
--> 136 checkpoint = torch.load(checkpoint)
137 model.load_state_dict(checkpoint['state_dict'])
139 if optimizer: