Ameet Deshpande

The label combinations are computed on the training set, so all the possible predictions will be some combination of labels seen during train time. So the short answer is that...

I agree that the name `pooler` might be a little confusing. The BERT model can be divided into three parts for understanding it easily 1. Embedding layer: Gets the embeddings...

Hi, are there any updates on this?

I just created a new flag called `all_boolean_args` and I pass all the boolean arguments as a comma separated list. For example, ``` bash --all_boolean_args do_train,do_eval,do_predict,overwrite_output_dir ``` In the code,...

Do you think it is possible to host the dataset for a while? I run into HTTP errors on some of the books, so this script ends up downloading only...