0.2.1-beta (first release)

First official release
v0.1.0 is skipped because this release is created after the initial commit.

Features:

• layers.Linear(): Feed-forward layer, applies weights a biases (optional) to the input.
• layers.Sequential(): A utility that is used to create a Sequential model.
• activations.ReLU(): Applies the ReLU activation to the input.
• activations.Softmax(): Applies the Softmax activation to the input.
• activations.Tanh(): Applies the Tanh activation to the input.
• loss.CrossEntropyLoss(): Calculates the Cross-Entropy loss of the predicted labels and true labels.
• loss.SparseCrossEntropyLoss(): Calculates the sparse Cross-Entropy loss of the predicted labels and true labels.

Note:
There may be discrepancies between the GitHub releases version and the PureTorch.__version__ during import. This is due to the initial custom versioning schema being updated to follow proper Semantic Versioning.