CHANGES
=======

* update setup.cfg
* remove PROJ\_SIZE in LSTM since it is incompatible with older version of Pytorch
* update crnn
* update generate attention mask
* split model to required interface and our default model
* move generate padding to utils
* update ConvAttnRNN to receive config and string decode only
* update conv\_seq2seq to receive Config and string decoder only
* add pad to Seq2SeqVocab
* fix leaf assign in Config
* restructure project
* remove branch docs from github actions
* refactor docs
* add ConvAttnRNN
* move generate mask to ocrstack/utils
* add log graph before training, fix logger name
* add tensorboard to github action
* add log\_scalars for logging multiple scalars at once; refactor logging format
* reset consolelogger's internal step to 0 when open
* add loggers
* move logger to engine package
* fix old loss imports
* add attention operations
* remove <eos> in prediction
* update constructor of ConsoleWriter to receive string instance
* fix console writer
* update trainer's test cases to work with both cpu and gpu
* fix lengths in GPU. It should be placed on CPU
* move rawText in metadata to required fields
* fix lengths in batch
* update visualize
* models now return loss
* add torchinfo to CI
* add torchinfo to summarize model
* move mask to device automatically
* update models
* temp
* add CSVDataset
* update init of loss functions
* add LabelSmoothLoss
* fix TransformerDecoder decode
* add option to remove <eos> when decode to string Seq2Seq
* fix loss end of sequence
* remove core package because it is redundant
* BatchPadTexts broadcast
* fix visualizer is None
* add Visualizer class
* update to build package
* add test attentionlstmdecoder
* add AttentionLSTMDecoder and refine Seq2SeqTrainBridge
* refactor
* refactor collate batch
* update metrics
* add CTCLoss wrapper
* fix metadata to device
* add metatdata field to Dataset
* fix transfer batch to device
* automatically use argmax when passing 3-dim tensor to CTC decode
* add type annotation for modelbase class
* add type annotation for collatebatch class
* add pyproject.toml
* update classifier
* add val\_step
* update trainer and its test script
* fix only run github actions on dev branch pushing
* add config comparison and its tests
* add build status badge to README
* remove incorrect imports
* add requirements
* remove redundant file
* fix flake8 errors
* update workflows
* update working\_directory in workflow githubaction
* debug workflow
* fix typo workflow
* only dev with py3.8 first
* refactor workflow
* fix typo
* update workflow to run test
* update workflow github actions
* add model seq2seq
* update tf sequence decoder
* add tests
* add Transformer Encoder Adapter
* add transformer decoder
* update index
* add sphinx extensions
* add data module to docs
* Update repo introduction
* Add readthedocs buiding status
* fail on warning = false
* add pyyaml to requirements
* add pillow, numpy, editdistance to requirements
* add torch and torchvision to requirements docs. We just need CPU version only
* add setup.py
* add module model
* add data module
* add config module
* add engine module
* add builder html option to readthedocs config
* add readthedocs configuration file
* add requirements.txt
* add branch to run actions
* add action for building docs
* add flake8 github actions
* add docs
* add typing
* add ocr metrics
* add image and string transforms
* add README
* add .flake8 config
* add gitignore
* add LICENSE
