Hiram Ring

A workaround that I found: install Open JDK Java 8 with brew using `brew cask install homebrew/cask-versions/adoptopenjdk8` (see https://github.com/Homebrew/homebrew-cask-versions/issues/7253#issuecomment-484672949) install Jenv with brew using `brew install jenv` and set your...

I'm not sure if this is the place to ask, but I'm wondering, given the state of the art for POS tagging as reported by: https://aclweb.org/aclwiki/POS_Tagging_(State_of_the_art) - is there a...

Very cool! I completely agree that active development is a key concern, along with speed, right up there with accuracy of POS tags. I simply wasn't aware of the spacy-stanfordnlp...

Interesting. It looks like the dependency parser doesn't handle conjoined clauses terribly well with a following 1st person pronoun. This is clearer with the raw text output: ``` print([x.text for...

You have to set up the `labels.csv` with multiple labels, and your `train.csv` and `val.csv` as described in the readme for multilabel classification. Then you need to set the `multilabel`...

+1 to this: would be nice to have a simple history callback that tracks the metrics etc, i.e. for use in visualizations.

when I train a multi-label classifier, `accuracy_thresh` outputs a different value than `accuracy_multilabel`. For example, the logged metrics return something like: ``` eval_accuracy_thresh after epoch 1: 0.876443088054657 eval_accuracy_multilabel after epoch...

just for clarification, freezing the layers means that there is a single linear layer (i.e. classifier head?) being trained for classification rather than all the layers being used to update...

aha - thanks! now that you point this out it makes sense, but it's not clear from the documentation. I spent an hour or so googling, and never came across...

this is related to the format of your datafiles, which can lead to issues when importing a CSV via a pandas dataframe. I might submit a pull request to allow...