Add PositionwiseFeedforward and TransformerDecoderLayer

Setup pyrefly

Impl MultiheadSelfAttention proj mats with linear module

Add cross-entropy loss function implementation

Add TransformerDecoderLayer module (not tested)

Add MultiheadSelfAttentionWithRoPE

Add MultiheadSelfAttention module; fix scaled dot product attention masking for multi dim batch

Add scaled dot product attention implementation

Add softmax function implementation and integrate into adapter tests

Add RotaryPositionalEmbedding module and integrate into adapter tests

Add RMSNorm module and integrate into adapter tests

Add Embedding module and integrate into adapter tests

Impl linear

Add VSCode settings for Python testing configuration

Fix BPETokenizer