tokenize | An open implementation of the Parse.com REST server | Runtime Evironment library

It looks like setuptools>=58 breaks support for use_2to3:

setuptools changelog for v58

So you should update setuptools to setuptools<58 or avoid using packages with use_2to3 in the setup parameters.

I was having the same problem, pip==19.3.1

The current release of pyodbc (4.0.32) does not have pre-built wheel files for Python 3.10. The easiest way to get it installed at the moment is to download the appropriate wheel from

https://www.lfd.uci.edu/~gohlke/pythonlibs/#pyodbc

and then install it. For example, if you are running 64-bit Python then you would download the 64-bit wheel and use

You could use a regex and extractall:

One of the dependencies for python-docx is lxml. The latest stable version of lxml is 4.6.3, released on March 21, 2021. On PyPI there is no lxml wheel for 3.10, yet. So it try to compile from source and for that Microsoft Visual C++ 14.0 or greater is required, as stated in the error.

However you can manually install lxml, before install python-docx. Download and install unofficial binary from Gohlke Alternatively you can use pipwin to install it from Gohlke. Note there may still be problems with dependencies for lxml.

Of course, you can also downgrade to python3.9.

EDIT: As of 14 Dec 2021 the latest lxml version 4.7.1 supports python 3.10

I am also stuck with the same problem because I installed the latest version of Node.js (v17.0.1).

Just go for node.js v14.18.1 and remove the latest version just use the stable version v14.18.1

It's not related to ES version.

Update max_expansions to more than 50.

max_expansions : Maximum number of variations created.

With 3 grams letter & digits as token_chars, ideal max_expansion will be (26 alphabets + 10 digits) * 3

1. removed /node_modules and yarn.lock file. 2.then reinstalled deleted packages/file using yarn install .

What you could do in this situation is to iterate on the validation set(or on the test set for that matter) and manually create a list of y_true and y_pred.

The ZeroShotClassificationPipeline is currently not supported by shap, but you can use a workaround. The workaround is required because:

1. The shap Explainer forwards only one parameter to the model (a pipeline in this case), but the ZeroShotClassificationPipeline requires two parameters, namely text, and labels.
2. The shap Explainer will access the config of your model and use its label2id and id2label properties. They do not match the labels returned from the ZeroShotClassificationPipeline and will result in an error.

Below is a suggestion for one possible workaround. I recommend opening an issue at shap and requesting official support for huggingface's ZeroShotClassificationPipeline.