lyhue1991 / eat_pytorch_in_20_days Goto Github PK

1-2图片建模的执行到features,labels = next(iter(dl_train))时报错_pickle.PicklingError: Can't pickle <function at 0x000002735CA54F28>: attribute lookup on main failed。是windows平台的问题吗

训练的时候报错。

划红线的部分，定义的时候与调用的时候，参数是不是写反了。不过是不是没影响？
位置在第 3-1,低阶API示范

6-2, 训练模型的3种方法
四，类风格 torchkeras.LightModel
line 33
acc = pl.metrics.functional.accuracy(preds, y)

报错如下：
AttributeError: module 'pytorch_lightning.metrics' has no attribute 'functional'

我是来学习pytorch的,看代码发现总是存在torchkeras这个库,如果这个库只是为了查看各层参数之类的.可以将其作为插件介绍一番,没必要篇篇文章都加入,感觉有点不太好.本来就是起步学习,真诚希望代码越简洁越好.

Failed to run summary...
Traceback (most recent call last):
File "D:\Program Files\pytorch'\1-2,图片数据建模流程范例.py", line 66, in
torchkeras.summary(net,input_shape= (3,32,32))
File "D:\Program Files\venv\lib\site-packages\torchkeras\summary.py", line 67, in summary
model(*x) if not (kwargs or args) else model(*x, *args, **kwargs)
File "D:\Program Files\venv\lib\site-packages\torch\nn\modules\module.py", line 1151, in _call_impl
hook_result = hook(self, input, result)
File "D:\Program Files\venv\lib\site-packages\torchkeras\summary.py", line 37, in hook
info["out"] = [batch_size] + list(outputs.size())[1:]
AttributeError: 'NoneType' object has no attribute 'size'
不使用summary后面也会出现AttributeError: 'NoneType' object has no attribute 'size'

trainer = pl.Trainer(max_epochs=20,gpus = 0, callbacks=[ckpt_cb])
trainer.fit(model,dl_train,dl_valid)
在运行这部分代码使用pytorch_lightning时总是报以下错误
'tqdm_notebook' object has no attribute 'reset'

你好，可以上传一下代码中的数据集吗？其中的covid-19.csv按照知乎的链接并没有找到对应的数据文件。请教一下！

一，准备数据 本文的数据集取自tushare，获取该数据集的方法参考了以下文章。
《https://zhuanlan.zhihu.com/p/109556102》

这个知乎链接失效了

when I click on the English links it just took me to the Chinese content

import torchtext的时候报错。
dlopen(/Users/Kingsun/opt/anaconda3/envs/py3/lib/python3.7/site-packages/torchtext/_torchtext.so, 2): Symbol not found: __ZNK3c104Type14isSubtypeOfExtERKNSt3__110shared_ptrIS0_EEPNS1_13basic_ostreamIcNS1_11char_traitsIcEEEE
Referenced from: /Users/Kingsun/opt/anaconda3/envs/py3/lib/python3.7/site-packages/torchtext/_torchtext.so
Expected in: /Users/Kingsun/opt/anaconda3/envs/py3/lib/python3.7/site-packages/torch/lib/libtorch_cpu.dylib
in /Users/Kingsun/opt/anaconda3/envs/py3/lib/python3.7/site-packages/torchtext/_torchtext.so

torchtext 的最新版本0.9.0取消了 Field，只有0.8.1有了。

nn.BCELoss输出未经过sigmod激活，nn.BCEWithLogitsLoss输出自动经过sigmod激活。
nn.CrossEntropyLoss 输出自动经过softmax激活

dl_train = DataLoader(TensorDataset(torch.tensor(x_train).float(),torch.tensor(y_train).float()),
shuffle = True, batch_size = 8)
dl_valid = DataLoader(TensorDataset(torch.tensor(x_test).float(),torch.tensor(y_test).float()),
shuffle = False, batch_size = 8)

AssertionError Traceback (most recent call last)
in
----> 1 dl_train = DataLoader(TensorDataset(torch.tensor(x_train).float(),torch.tensor(y_train).float()),
2 shuffle = True, batch_size = 8)
3 dl_valid = DataLoader(TensorDataset(torch.tensor(x_valid).float(),torch.tensor(y_valid).float()),
4 shuffle = False, batch_size = 8)

~\anaconda3\envs\pytorch\lib\site-packages\torch\utils\data\dataset.py in init(self, *tensors)
34
35 def init(self, *tensors):
---> 36 assert all(tensors[0].size(0) == tensor.size(0) for tensor in tensors)
37 self.tensors = tensors
38

AssertionError:

scalar = torch.tensor(True)
print(scalar.size())
print(vector.shape)

scalar = torch.tensor(True)
print(scalar.size())
print(scalar.shape)

dl_train = DataLoader(TensorDataset(torch.tensor(x_train).float(),torch.tensor(y_train).float()),
shuffle = True, batch_size = 8)
dl_val = DataLoader(TensorDataset(torch.tensor(x_test).float(),torch.tensor(y_test).float()),
shuffle = False, batch_size = 8)

TypeError Traceback (most recent call last)
Cell In[9], line 1
----> 1 dl_train = DataLoader(TensorDataset(torch.tensor(x_train).float(),torch.tensor(y_train).float()),
2 shuffle = True, batch_size = 8)
3 dl_val = DataLoader(TensorDataset(torch.tensor(x_test).float(),torch.tensor(y_test).float()),
4 shuffle = False, batch_size = 8)

TypeError: can't convert np.ndarray of type numpy.object_. The only supported types are: float64, float32, float16, complex64, complex128, int64, int32, int16, int8, uint8, and bool.

请问torchkeras这个模块引用时出错，name 'torchkeras' is not defined，
是自己写的吗？在哪里可以找到？谢谢

in1-2 code:
for x,y in dl_trian:
print(x.shape,y.shape)
break
error:_pickle.PicklingError: Can't pickle <function at 0x000002DDC0CBB048>: attribute lookup on main failed.
because the error the whole code can run correctly

前面二分类问题可以通过加入参数来解决报错：metric_dict = {"acc":Accuracy(task="binary")}。
但是碰到多分类问题，改成metric_dict = {"acc":Accuracy(task="multiclass")}之后，就会报新错。

另外我发现这个项目里用到Accuracy的地方作者都没有加task的参数，不知道是不是这个包版本更新的问题？

ipython-input-5-ad41ec68ba80> in
7 plt.figure(figsize=(8,8))
8 for i in range(9):
----> 9 img,label = ds_train[i]
10 img = img.permute(1,2,0)
11 ax = plt.subplot(3,3,i+1)

E:\anaconda\envs\pytorch\lib\site-packages\torchvision\datasets\folder.py in getitem(self, index)
151 sample = self.loader(path)
152 if self.transform is not None:
--> 153 sample = self.transform(sample)
154 if self.target_transform is not None:
155 target = self.target_transform(target)

E:\anaconda\envs\pytorch\lib\site-packages\torchvision\transforms\transforms.py in call(self, img)
65 def call(self, img):
66 for t in self.transforms:
---> 67 img = t(img)
68 return img
69

TypeError: object() takes no parameters

作者考虑添加使用 torch.distributed 加速多GPU并行训练的内容吗？貌似比nn.DataParallel快很多

class Model(torchkeras.LightModel):
AttributeError: module 'torchkeras' has no attribute 'LightModel'