How to convert TorchText 0.9 to the latest version about transformer HOT 10 OPEN

i used conda to install torchtext, and it seems that it recommends torchtext==0.6.0, then you actually don't need to change anything but the two lines in util/data_loader(line 6, 7) and it works well!

Hello, my friend, can you give me a copy of your requirements.txt? I would be very grateful.

python3.8
torch==1.7
torchvison==0.8.1
torchtext==0.8.0

replace the

from torchtext.legacy.data import Field, BucketIterator
from torchtext.legacy.datasets import Multi30k

by

from torchtext.data import Field, BucketIterator
from torchtext.datasets import Multi30k

if the data can't dowload, you can download them from

https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/training.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/validation.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/mmt16_task1_test.tar.gz'

and tar them in the .data/multi30k folder

from transformer.

i used conda to install torchtext, and it seems that it recommends torchtext==0.6.0,

then you actually don't need to change anything but the two lines in util/data_loader(line 6, 7)

and it works well!

from transformer.

i used conda to install torchtext, and it seems that it recommends torchtext==0.6.0, then you actually don't need to change anything but the two lines in util/data_loader(line 6, 7) and it works well!

Hei, bro, could you pls tell me how to download the datasets, thanks in advance!

from transformer.

I‘m new to transformer rencently,too. My torchtext==0.15, the fuction Field and Bucketerator were removed . What are the alternatives to torchtext.field and Buckerator?

from transformer.

i used conda to install torchtext, and it seems that it recommends torchtext==0.6.0, then you actually don't need to change anything but the two lines in util/data_loader(line 6, 7) and it works well!

Hello, my friend, can you give me a copy of your requirements.txt? I would be very grateful.

from transformer.

i used conda to install torchtext, and it seems that it recommends torchtext==0.6.0, then you actually don't need to change anything but the two lines in util/data_loader(line 6, 7) and it works well!

Hello, my friend, can you give me a copy of your requirements.txt? I would be very grateful.

python3.8 torch==1.7 torchvison==0.8.1 torchtext==0.8.0

replace the

from torchtext.legacy.data import Field, BucketIterator
from torchtext.legacy.datasets import Multi30k

by

from torchtext.data import Field, BucketIterator
from torchtext.datasets import Multi30k

if the data can't dowload, you can download them from

https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/training.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/validation.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/mmt16_task1_test.tar.gz'

and tar them in the .data/multi30k folder

okay, thank u, my friends!

from transformer.

我使用 conda 安装 torchtext，似乎它推荐 torchtext==0.6.0，那么您实际上不需要更改任何东西，只需更改 util/data_loader（第 6、7 行）中的两行，效果很好！

你好，我的朋友，你能给我一份你的要求.txt吗？我将不胜感激。

python3.8 火炬==1.7 火炬视子==0.8.1 火炬文本==0.8.0

将

from torchtext.legacy.data import Field, BucketIterator
from torchtext.legacy.datasets import Multi30k

由

from torchtext.data import Field, BucketIterator
from torchtext.datasets import Multi30k

如果数据无法下载，您可以从

https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/training.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/validation.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/mmt16_task1_test.tar.gz'

并将它们涂在 .data/multi30k 文件夹中

您好 我按照您后面的方法 将三个文件下载并放到文件夹.data/multi30k 中，但是依然提示缺少train.en文件 请问还有其他的办法下载数据或者有下载脚本吗

from transformer.

我使用 conda 安装 torchtext，似乎它推荐 torchtext==0.6.0，那么您实际上不需要更改任何东西，只需更改 util/data_loader（第 6、7 行）中的两行，效果很好！

你好，我的朋友，你能给我一份你的要求.txt吗？我将不胜感激。

python3.8 火炬==1.7 火炬视子==0.8.1 火炬文本==0.8.0
将

from torchtext.legacy.data import Field, BucketIterator
from torchtext.legacy.datasets import Multi30k

由

from torchtext.data import Field, BucketIterator
from torchtext.datasets import Multi30k

如果数据无法下载，您可以从

https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/training.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/validation.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/mmt16_task1_test.tar.gz'

并将它们涂在 .data/multi30k 文件夹中

您好 我按照您后面的方法 将三个文件下载并放到文件夹.data/multi30k 中，但是依然提示缺少train.en文件 请问还有其他的办法下载数据或者有下载脚本吗

没解压吧，解压一下

from transformer.

我使用 conda 安装 torchtext，似乎它推荐 torchtext==0.6.0，那么您实际上不需要更改任何东西，只需更改 util/data_loader（第 6、7 行）中的两行，效果很好！

你好，我的朋友，你能给我一份你的要求.txt吗？我将不胜感激。

python3.8 火炬==1.7 火炬视子==0.8.1 火炬文本==0.8.0
将

from torchtext.legacy.data import Field, BucketIterator
from torchtext.legacy.datasets import Multi30k

由

from torchtext.data import Field, BucketIterator
from torchtext.datasets import Multi30k

如果数据无法下载，您可以从

https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/training.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/validation.tar.gz'
https://raw.githubusercontent.com/neychev/small_DL_repo/master/datasets/Multi30k/mmt16_task1_test.tar.gz'

并将它们涂在 .data/multi30k 文件夹中

您好 我按照您后面的方法 将三个文件下载并放到文件夹.data/multi30k 中，但是依然提示缺少train.en文件 请问还有其他的办法下载数据或者有下载脚本吗

没解压吧，解压一下

解决了我的问题，非常感谢！！！

from transformer.

I‘m new to transformer rencently,too. My torchtext==0.15, the fuction Field and Bucketerator were removed . What are the alternatives to torchtext.field and Buckerator?我最近也是变压器新手。我的torchtext==0.15，功能Field和Bucketerator被删除了。 torchtext.field 和 Buckerator 的替代品有哪些？

I have this problem,too.Do you solve this problem?

from transformer.