neuralnetlib 4.3.4

• feat+fix(gan+compiles): metrics in compile + fixed labels in GAN
• refactor(gan): evaluate
• fix(gan): metrics handling
• fix(mmd): numerical instability
• refactor(gan): some improvements
• fix(dropout): input shape for gans
• feat(gan): add adaptive plot sample gen according to activation function
• docs(notebook): update gan example
• docs(readme): update transformer description
• fix(cgan): conditional handling in G/D
• fix(lr): add gan support
• feat: add balanced batch sampling
• fix(docs): notebook example for gan
• build: correct version
• fix: beam search

neuralnetlib 4.3.3

• fix(configs): layers states saving and loading
• fix(configs): losses states saving and loading
• fix(configs): optimizers states saving and loading
• fix(configs): models states saving and loading
• ci: bump version to 4.3.3

neuralnetlib 4.3.2

• docs: update readme
• fix(dropout/batchnorm): save&load
• ci: bump version to 4.3.2

neuralnetlib 4.3.1

• fix(batchnorm): init
• ci: bump version to 4.3.1

neuralnetlib 4.3.0

• fix(git): file caching
• feat: add support for multi-label classification
• fix(cgan): y_train when None
• feat(cgan): add label smoothing
• fix(cgan): predictions
• docs(gans): update
• docs(notebook): add svm
• feat(utils): add make_blobs
• feat(utils): add make_classification
• feat(metrics): add adjusted_rand_score
• feat(metrics): add adjusted_mutual_info_score
• fix(metrics): MAE, MSE, MAPE
• refactor: usage of standalone functions everywhere
• ci: bump version to 4.3.0

neuralnetlib 4.2.0

• docs(notebook): fresh run²
• docs(readme): update quick examples
• docs(readme): update²
• docs: add conv example for gan
• feat(layer): add Conv2DTranspose
• docs: add conv example for gan using conv2dtranspose
• feat(GAN): add Conditional GAN (CGAN)
• ci: bump version to 4.2.0

neuralnetlib 4.1.0

• fix(LSTM): huge improvements in gradient flow
• feat(gradient_norm): better batch handling
• fix(Callbacks): now fully working generically
• fix(metrics): val_loss
• fix(Embedding): bias in output
• fix(display): val_*
• feat(metric): add pearsonr
• feat(metric): add kurtosis
• feat(metric): add skew
• fix(metric): skewness and kurtosis when variance=0
• fix(gan): batch logs
• fix(activation): config loading
• feat(lstm): improved flexibility and parameterizing
• docs(notebook): fresh run
• ci: bump version to 4.1.0

neuralnetlib 4.0.7

• fix(LSTM): huge improvements in gradient flow
• feat(gradient_norm): better batch handling
• fix(Callbacks): now fully working generically
• fix(metrics): val_loss
• fix(Embedding): bias in output
• fix(display): val_*
• feat(metric): add pearsonr
• feat(metric): add kurtosis
• feat(metric): add skew
• fix(metric): skewness and kurtosis when variance=0
• fix(gan): batch logs
• fix(activation): config loading
• feat(lstm): improved flexibility and parameterizing
• docs(notebook): fresh run
• ci: bump version to 4.0.7

neuralnetlib 4.0.6

• refactor: autopep8 format
• ci: bump version to 4.0.6

neuralnetlib 4.0.5

• fix(cosine_sim): division by zero
• refactor: remove the use of enums
• docs: huge update and new examples
• docs(readme): update hyperlinks
• feat: add SVM
• fix(Dense): bad variable init
• fix(Tokenizer): bpe tokenizing
• fix: jacobian matrix instead of approximation
• fix(AddNorm): backward pass
• ci: bump version to 4.0.4