Order the steps of how the Transformer architecture processes input data from initial encoding to final output generation.

First the raw text is split into tokens, each token is turned into a vector by an embedding layer and a positional encoding is added to give order information. These vectors pass through several encoder blocks, each using self‑attention to mix information from all positions and a feed‑forward network to refine the representations. The resulting encoder output is then fed to the decoder, which first attends to itself using self‑attention on the already generated output. Next the decoder attends to the encoder output to incorporate context from the input. Finally a linear layer followed by a softmax produces the probability of each next word, and the most likely word is chosen to continue the sequence.