Dear TF Hub Team,

USE paper Section 5 has a interesting paragraph on evaluation where authors use Arc Cosine (Cos Inverse) whose range is 0 to Pi in radians instead of plain cosine distance with range 0 to 2.

". For the pairwise semantic similarity task, we directly assess
the similarity of the sentence embedding produced by our two encoders. As show

Expected Behavior

I want to convert torch.nn.Linear modules to weight drop linear modules in my model (possibly big), and I want to train my model with multi-GPUs. However, I have RuntimeError in my sample code. First, I have _weight_drop() which drops some part of weights in torch.nn.Linear (see the code below).

Actual Behavior

RuntimeError: arguments are located on different GPUs at /

(tensorflow) F:\Postgraduate\KaggleLearning\multi-class-text-classification-cnn-rnn-master\multi-class-text-classification-cnn-rnn-master>python predict.py ./t
rained_results_1541818386/ ./data2/samples.csv
D:\Anaconda\anaconda\envs\tensorflow\lib\site-packages\gensim\utils.py:1212: UserWarning: detected Windows; aliasing chunkize to chunkize_serial
warnings.warn("detected Windows; aliasing c

What code should be added so that the word embedding and confusion matrix at each step could be visualized in tensorboard?

I viewed the whole code and found that the code only use toy dummy data to train model. So I don't really understand how you use those data to train GCN model. Can you supply the code or instructions about how to use real-world data to train model?

Loading pre-trained word wiki.txt.word_vectors from the location - 'http://ltdata1.informatik.uni-hamburg.de/sensegram/en/'. Some of the words do not have 300 dimensions. Loading it using gensim package gives an error.