Merge branch 'master' into 'pytorch-lightning-models'

# Conflicts:
#   .idea/combolightning.iml
#   README.md
#   combo/common/util.py
#   combo/data/__init__.py
#   combo/data/api.py
#   combo/data/dataset_readers/dataset_reader.py
#   combo/data/dataset_readers/utils.py
#   combo/data/fields/__init__.py
#   combo/data/fields/adjacency_field.py
#   combo/data/fields/field.py
#   combo/data/fields/sequence_label_field.py
#   combo/data/fields/sequence_multilabel_field.py
#   combo/data/fields/text_field.py
#   combo/data/instance.py
#   combo/data/samplers/batch_sampler.py
#   combo/data/token_indexers/__init__.py
#   combo/data/token_indexers/pretrained_transformer_fixed_mismatched_indexer.py
#   combo/data/token_indexers/pretrained_transformer_indexer.py
#   combo/data/token_indexers/pretrained_transformer_mismatched_indexer.py
#   combo/data/token_indexers/token_characters_indexer.py
#   combo/data/token_indexers/token_const_padding_characters_indexer.py
#   combo/data/token_indexers/token_features_indexer.py
#   combo/data/token_indexers/token_indexer.py
#   combo/data/tokenizers/__init__.py
#   combo/data/tokenizers/character_tokenizer.py
#   combo/data/tokenizers/pretrained_transformer_tokenizer.py
#   combo/data/tokenizers/spacy_tokenizer.py
#   combo/data/tokenizers/tokenizer.py
#   combo/data/tokenizers/whitespace_tokenizer.py
#   combo/data/vocabulary.py
#   combo/main.py
#   combo/models/__init__.py
#   combo/models/archival.py
#   combo/models/combo_model.py
#   combo/models/encoder.py
#   combo/modules/augmented_lstm.py
#   combo/modules/encoder.py
#   combo/modules/input_variational_dropout.py
#   combo/nn/__init__.py
#   combo/nn/regularizers/__init__.py
#   combo/nn/regularizers/regularizers.py
#   combo/predict.py
#   combo/predictors/predictor.py
#   combo/training/__init__.py
#   combo/training/scheduler.py
#   combo/utils/__init__.py
#   combo/utils/file_utils.py
#   combo/utils/graph.py
#   requirements.txt
#   tests/data/fields/test_samplers.py
#   tests/data/test_vocabulary.py