torchtext legacy

[g-i-o-r-g-i-o]

Hello, how do I run this in google colab? The problem is that data and datasets have now moved in .legacy. thanks for any help

01_rnn_simple_imdb.ipynb

same problem in

1_lstm.ipynb

### Defining the feature processing

TEXT = torchtext.legacy.data.Field(
    tokenize='spacy', # default splits on whitespace
    tokenizer_language='en_core_web_sm'
)

### Defining the label processing

LABEL = torchtext.legacy.data.LabelField(dtype=torch.long

---

AttributeError Traceback (most recent call last) in 1 ### Defining the feature processing 2 ----> 3 TEXT = torchtext.legacy.data.Field( 4 tokenize='spacy', # default splits on whitespace 5 tokenizer_language='en_core_web_sm'

AttributeError: module 'torchtext' has no attribute 'legacy'

[rasbt]

You can try to downgrade torchtext to an older version. E.g., torchtext 0.9 should support torchtext.legacy. To downgrade it, you can run

!pip install torchtext==0.9

in a Google Colab cell