practicingman / chinese_text_cnn Goto Github PK

有人遇到这个问题吗？感谢

Hi, could you tell me where is the dataset comes from?

在高版本的Pytorch（我的1.5.0）中train.py中的feature.data.t_(), target.data.sub_(1)会报错，改成以下就没有问题了：

feature = feature.data.t()
target = target.data.sub(1)

您好！
关于您在执行multichannel模式时，命令行选项是
python main.py -static=true -non-static=true -multichannel=true
对应代码是

if args.static:
    self.embedding = self.embedding.from_pretrained(args.vectors, freeze=not args.non_static)
if args.multichannel:
    self.embedding2 = nn.Embedding(vocabulary_size, embedding_dimension).from_pretrained(args.vectors)

当三个选项均为True时，freeze=not True即freeze=False, 即embedding是微调了的。同时embedding2也是微调的。那这样不就是两个都微调了吗？？？
是不是应该设置 -non-static=False?

Loading data...
The device argument should be set by using torch.device or passing a string
as an argument. This behavior will be deprecated soon and currently defaults to
cpu.
The device argument should be set by using torch.device or passing a string
as an argument. This behavior will be deprecated soon and currently defaults to
cpu.
Parameters:
BATCH_SIZE=128
CLASS_NUM=3
CUDA=False
DEVICE=-1
DROPOUT=0.5
EARLY_STOPPING=1000
EMBEDDING_DIM=128
EPOCHS=256
FILTER_NUM=100
FILTER_SIZES=[3, 4, 5]
LOG_INTERVAL=1
LR=0.001
MAX_NORM=3.0
MULTICHANNEL=False
NON_STATIC=False
PRETRAINED_NAME=sgns.zhihu.word
PRETRAINED_PATH=pretrained
SAVE_BEST=True
SAVE_DIR=snapshot
SNAPSHOT=None
STATIC=False
TEST_INTERVAL=100
VOCABULARY_SIZE=35572
Traceback (most recent call last):
File "main.py", line 98, in
train.train(train_iter, dev_iter, text_cnn, args)
File "d:\我的文档\桌面\workspace2\chinese_text_cnn\train.py", line 18, in trai
n
feature.data.t_(), target.data.sub_(1)
RuntimeError: set_storage_offset is not allowed on a Tensor created from .data o
r .detach().
If your intent is to change the metadata of a Tensor (such as sizes / strides /
storage / storage_offset)
without autograd tracking the change, remove the .data / .detach() call and wrap
the change in a with torch.no_grad(): block.
For example, change:
x.data.set_(y)
to:
with torch.no_grad():
x.set_(y)

请问怎样做单条预测

求大神帮忙解答一下，我用您的程序在我电脑上运行，总是会报这个错
Traceback (most recent call last):
File "E:/software/code/sentimentanalysis/chinese_text_cnn/main.py", line 70, in
train_iter, dev_iter = load_dataset(text_field, label_field, args, device=-1, repeat=False, shuffle=True)
File "E:/software/code/sentimentanalysis/chinese_text_cnn/main.py", line 52, in load_dataset
train_dataset, dev_dataset = dataset.get_dataset('data', text_field, label_field)
File "E:\software\code\sentimentanalysis\chinese_text_cnn\dataset.py", line 23, in get_dataset
('text', text_field)
File "C:\Users\Administrator\Anaconda3\lib\site-packages\torchtext\data\dataset.py", line 78, in splits
os.path.join(path, train), **kwargs)
File "C:\Users\Administrator\Anaconda3\lib\site-packages\torchtext\data\dataset.py", line 269, in init
next(reader)
File "C:\Users\Administrator\Anaconda3\lib\site-packages\torchtext\utils.py", line 130, in unicode_csv_reader
csv.field_size_limit(sys.maxsize)
OverflowError: Python int too large to convert to C long
我的电脑系统是win10，python版本是python3.6

Hi I ran the repo on both google colab and windows 10, got same error both times:

help please.

在训练阶段调用完测试函数后（https://github.com/bigboNed3/chinese_text_cnn/blob/master/train.py#L37）是否应该添加一行model.train()恢复到训练模式呀？

首先非常感谢作者您分享的代码，只是如何编写预测代码呢，我这里不是太懂

Traceback (most recent call last):
File "main.py", line 70, in
train_iter, dev_iter = load_dataset(text_field, label_field, args, device=-1, repeat=False, shuffle=True)
File "main.py", line 57, in load_dataset
text_field.build_vocab(train_dataset, dev_dataset)
File "/home/mli/.pyenv/versions/miniconda-latest/envs/li/lib/python3.7/site-packages/torchtext/data/field.py", line 298, in build_vocab
for x in data:
File "/home/mli/.pyenv/versions/miniconda-latest/envs/li/lib/python3.7/site-packages/torchtext/data/dataset.py", line 154, in getattr
yield getattr(x, attr)
AttributeError: 'Example' object has no attribute 'text'

求大神解答

报错：
RuntimeError: Expected 4-dimensional input for 4-dimensional weight [100, 1, 3, 128], but got 3-dimensional input of size [2495, 1, 128] instead

输出size
Batch[9] - loss: 0.666672 acc: 86.7188%(111/128)1 torch.Size([128, 1, 103, 128])
torch.Size([128, 6])
Batch[10] - loss: 0.486642 acc: 90.6250%(116/128)1 torch.Size([2495, 1, 128])

明明是二分类的数据，为何显示class_num=3呢?

你好，chinese_text_cnn/model.py中的第20行中的self.embedding.from_pretrained这个函数在哪里？好像没有找到定义的地方？