深度学习框架PyTorch一书的学习-第六章-实战指南 Log

 

目录

深度学习框架PyTorch一书的学习-第六章-实战指南

python命令：

cd E:\DeepingLearn\DogvsCat\pytorch-best-practice

python main.py train --train-data-root=./data/train --use-gpu --env=classifier

 

AttributeError: 'dict' object has no attribute 'iteritems'

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

AssertionError: Torch not compiled with CUDA enabled

 

IndexError: invalid index of a 0-dim tensor. Use tensor.item() to convert a 0-dim tensor to a Python number

NameError: name 'unicode' is not defined

---

深度学习框架PyTorch一书的学习-第六章-实战指南

https://github.com/chenyuntc/pytorch-book/tree/v1.0/chapter6-实战指南

Pytorch学习遇到的问题记录

python命令：

cd E:\DeepingLearn\DogvsCat\pytorch-best-practice
python main.py train --train-data-root=./data/train --use-gpu --env=classifier

 

AttributeError: 'dict' object has no attribute 'iteritems'

(base) E:\DeepingLearn\DogvsCat\pytorch-best-practice>python main.py train --train-data-root=./data/train --use-gpu --env=classifier
Traceback (most recent call last):
  File "main.py", line 171, in <module>
    fire.Fire()
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 138, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 471, in _Fire
    target=component.__name__)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 675, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "main.py", line 49, in train
    opt.parse(kwargs)
  File "E:\DeepingLearn\DogvsCat\pytorch-best-practice\config.py", line 30, in parse
    for k,v in kwargs.iteritems():
AttributeError: 'dict' object has no attribute 'iteritems'

 

【解决方法】config.py中把 ‘iteritems' 改成 items。最好使用python2，best-practice 当初是用python2 写的

 

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

Traceback (most recent call last):
  File "main.py", line 171, in <module>
    fire.Fire()
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 138, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 471, in _Fire
    target=component.__name__)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 675, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "main.py", line 55, in train
    model.load(opt.load_model_path)
  File "E:\DeepingLearn\DogvsCat\pytorch-best-practice\models\BasicModule.py", line 19, in load
    self.load_state_dict(t.load(path))
  File "D:\Anaconda3\lib\site-packages\torch\serialization.py", line 381, in load
    f = open(f, 'rb')
FileNotFoundError: [Errno 2] No such file or directory: 'checkpoints/model.pth'

【解决方法】把config.py中的load_model_path='checkpoints/model.pth'改成load_model_path=None

 

AssertionError: Torch not compiled with CUDA enabled

Traceback (most recent call last):
  File "main.py", line 171, in <module>
    fire.Fire()
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 138, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 471, in _Fire
    target=component.__name__)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 675, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "main.py", line 56, in train
    if opt.use_gpu: model.cuda()
  File "D:\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 311, in cuda
    return self._apply(lambda t: t.cuda(device))
  File "D:\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 208, in _apply
    module._apply(fn)
  File "D:\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 208, in _apply
    module._apply(fn)
  File "D:\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 230, in _apply
    param_applied = fn(param)
  File "D:\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 311, in <lambda>
    return self._apply(lambda t: t.cuda(device))
  File "D:\Anaconda3\lib\site-packages\torch\cuda\__init__.py", line 178, in _lazy_init
    _check_driver()
  File "D:\Anaconda3\lib\site-packages\torch\cuda\__init__.py", line 92, in _check_driver
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

【解决方法】由于电脑无GPU加速,在config.py中更改use_gpu = False.

python main.py train --train-data-root=./data/train --env=classifier

【Log】电脑无cuda，能否利用核显加速？

 

IndexError: invalid index of a 0-dim tensor. Use tensor.item() to convert a 0-dim tensor to a Python number

raceback (most recent call last):
  File "main.py", line 171, in <module>
    fire.Fire()
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 138, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 471, in _Fire
    target=component.__name__)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 675, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "main.py", line 99, in train
    loss_meter.add(loss.data[0])
IndexError: invalid index of a 0-dim tensor. Use tensor.item() to convert a 0-dim tensor to a Python number

【解决方法】

报错原因分析:

train_loss += loss.data[0] 是pytorch0.3.1版本代码,在0.4-0.5版本的pytorch会出现警告,不会报错,但是0.5版本以上的pytorch就会报错,总的来说是版本更新问题.

解决方法:

#将原语句：

train_loss+=loss.data[0]

#修改为：

train_loss+=loss.item()

从tensor中取值，新版pytorch要使用.item()，应该是这个原因。在chap6的main.py中修改：

修改前：

# meters update and visualize
loss_meter.add(loss.data[0])
 confusion_matrix.add(score.data, target.data)

修改后：

 # meters update and visualize
loss_meter.add(loss.item())
confusion_matrix.add(score.data, target.data)

 

NameError: name 'unicode' is not defined

raceback (most recent call last):
  File "main.py", line 171, in <module>
    import fire
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 138, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 471, in _Fire
    target=component.__name__)
  File "D:\Anaconda3\lib\site-packages\fire\core.py", line 675, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "main.py", line 103, in train
    vis.plot('loss', loss_meter.value()[0])
  File "E:\DeepingLearn\DogvsCat\pytorch-best-practice\utils\visualize.py", line 44, in plot
    win=unicode(name),
NameError: name 'unicode' is not defined

【解决方法】把# coding:utf8

改为# encoding: utf-8

或# -*- coding: utf-8 -*-

 

或直接删掉这一行

原因待查

unicode(name)改为str(name）

#修改前
self.vis.line(Y=np.array([y]), X=np.array([x]),
                      win=unicode(name),
                      opts=dict(title=name),
                      update=None if x == 0 else 'append',
                      **kwargs
                      )
#修改后

self.vis.line(Y=np.array([y]), X=np.array([x]),
                      win=str(name),
                      opts=dict(title=name),
                      update=None if x == 0 else 'append',
                      **kwargs
                      )