prdwb / bert_hae Goto Github PK

TypeError: convert_examples_to_variations_and_then_features() missing 2 required positional arguments: 'glove' and 'tfidf_vectorizer'

Hi, I ran the script and got evaluation results like this:

epoch finished!
evaluation: 24000, total_loss: 1.491316556930542, f1: 59.33490221751538, followup: 0.0, yesno: 19.3518941122775, heq: 55.84968811805872, dheq: 4.5

Model saved in path OUTPUT_DIR/model_24000.ckpt

The paper shows F1=63.1/62.4

thank you!

It seems like its taking several days for the training to be completed. Is that supposed to be the case?

When I ran the hae.py file, after about 20 mins, I got this:
Traceback (most recent call last):
File "hae.py", line 104, in
with open(features_fname, 'wb') as handle:
FileNotFoundError: [Errno 2] No such file or directory: '/home/lef/Desktop/bert_output/cache/quac/train_features_False_11.pkl'.

There was no genration of pkl files in the cache folder. I have checked the dir, and there was nothing wrong. Wondering why this happened.

I am running this with python3.7, cuda 10 and tf-gpu 1.13.

Hi !
First, thank you for this great model! i'am wondering how can i add early stopping support in your model? I have a training set, a validation set and a test set, I want to train on the training set, do early stopping with the validation set and get predictions on the test set.

How can I fix it? thanks a lot

why you comment out FLAGS.history, and what's the different between FLAGS.history and FLAGS.max_considered_history_turns