Closed rcalsaverini closed 5 years ago
Since BERTMaskedLM
is a model, you need to call .build()
on it before trying to use it so the Docker image is built. i.e.:
from gobbli.augment import Word2Vec, BERTMaskedLM
import pathlib
wv = BERTMaskedLM(data_dir=pathlib.Path("./data/"))
wv.build()
X = ["Soccer is a fun game."]
wv.augment(X, times=10, p=0.5)
We really should be catching the ImageNotFound
error and give you a more helpful error message, though. I'll see about changing that.
Fixed in new release (0.0.3). You should now see this error if you try this:
RuntimeError: gobbli couldn't find the Docker image for the container it was asked to run. This probably means you didn't call .build() on a model before using it.
Thanks @jasonnance. Doing this solved the problem.
System Information
Description
I tried to run the
BERTMaskedLM
augmentation and got the following error:ImageNotFound: 404 Client Error: Not Found ("pull access denied for gobbli-bert-maskedlm, repository does not exist or may require 'docker login': denied: requested access to the resource is denied")
Code for Minimal Reproducible Example
Install gobbli in a fresh virtualenvironment via
pip install gobbli
.Run the following code:
Output
This is the full error output: