Added - Support for additional model architectures: RoBERTa, DeBERTa - Documentation: reproduction of results
Changed - Renamed class: Experiment -> Training - Renamed training parameters: prune_ratio_train -> train_fraction (+ same for val & test)
Fixed - Controlled exception in case of multiple GPUs ([6](https://github.com/flxst/nerblackbox/pull/6)) - Download data from LabelStudio ([9](https://github.com/flxst/nerblackbox/pull/9), [#10](https://github.com/flxst/nerblackbox/pull/10))
Changed - restructured docs - reduced CLI to two commands (nerblackbox mlflow & nerblackbox tensorboard) - dropped support for python version 3.11 - upgraded dependencies (fixing potential security vulnerabilities)
0.0.14
Not secure
Added - Model: prediction on file - Model: evaluation of any model on any dataset
Changed - API: complete renewal using classes Store, Dataset, Experiment, Model - Supported python versions: 3.8 to 3.11 - Dataset: no shuffling by default
Fixed - Model: base model with NER classification head can be loaded
0.0.13
Not secure
Added - NerModelPredict: GPU batch inference - TextEncoder class for custom data preprocessing - HuggingFace datasets integration: enable subsets - HuggingFace datasets: support for sucx_ner
Changed - NerModelPredict: improved inference time and data post-processing - API: load best model of experiment directly (instead of via ExperimentResults) - upgrade pytorch-lightning
0.0.12
Not secure
Added - Adaptive fine-tuning - Integration of HuggingFace Datasets - Integration of raw (unpretokenized) data - Integration of different annotation schemes and seamless conversion between them - Option to specify experiments dynamically (instead of using a config file) - Option to add special tokens - New built-in dataset: Swe-NERC - Use seeds for reproducibility
Changed - Validation only on single metric (e.g. loss) during training - Shuffling of all datasets (train, val, test) - Results: epochs start counting from 1 instead of 0 - Results: compute standard version of macro-average, plus number of contributing classes - Results: add precision and recall
Fixed - All models that are based on WordPiece tokenizer work - Early stopping: use last model instead of stopped epoch model
0.0.11
Not secure
Added - NerModelPredict: predict on token or entity level - Evaluation entity level: compute metrics for single labels - Evaluation token level: confusion matrix - Evaluation token & entity level: number of predicted classes
Changed - Evaluation token level: use plain annotation scheme - Migrate to pytorch-lightning==1.3.7, seqeval==1.2.2, mlflow==1.8.0