Recurrent Neural Networks (RNNs) have been a staple in sequence modeling and machine translation. However, the inherent sequential computation in RNNs limits parallelization, creating inefficiencies. Even with the integration of attention mechanisms, RNNs still depend on sequential processing, which restricts potential performance enhancements. Proposed Model: Transformer The Transformer model introduces a novel approach to sequence modeling by utilizing self-attention mechanisms instead of recurrence. This model completely removes the need for recurrence and convolutions, relying entirely on encoder-decoder attention mechanisms to achieve better performance and efficiency. Model Architecture The Transformer architecture consists of stacks of encoders and decoders, each comprising multiple identical layers. Each layer includes two primary sub-layers: 1. Self-Attention Mechanism 2. Fully Connected Feed-Forward Network Residual connections and layer normalization are applied around each sub-layer to ensure stability and efficiency during training.