Emotion Classification using BERT

Evaluation result

INFO:tensorflow: eval_accuracy = 0.65049815

INFO:tensorflow: eval_loss = 1.3037859

INFO:tensorflow: global_step = 1908

INFO:tensorflow: loss = 1.3041745

If you want to see the predicted result then go to the notebook file

The dataset are just the sample of original dataset.

tensorflow >= 1.12

python >= 3.5

urllib, urllib3 == latest

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
src_tokenizer		src_tokenizer
README.md		README.md
dev_sample.tsv		dev_sample.tsv
modeling.py		modeling.py
optimization_multi_gpu.py		optimization_multi_gpu.py
preprocess emotion datset.ipynb		preprocess emotion datset.ipynb
run_classifier_ETRI.sh		run_classifier_ETRI.sh
run_classifier_ETRI_multigpu.py		run_classifier_ETRI_multigpu.py
run_classifier_ETRI_multigpu_hook.py		run_classifier_ETRI_multigpu_hook.py
run_classifier_ETRI_multigpu_savepb.py		run_classifier_ETRI_multigpu_savepb.py
run_classifier_ETRI_predict.sh		run_classifier_ETRI_predict.sh
test.tsv		test.tsv
train_sample.tsv		train_sample.tsv