LLMs From Scratch Day 8

I'm a day late posting this, but yesterday I put the end-to-end model together and successfully completed a forward pass. This was actually relatively easy to do, and mostly amounted to writing functions to create a dictionary with each layer of the encoder and decoder sequentially, along with the embedding layer. Then the forward pass was just a matter of iterating through the layers. Tomorrow will be more interesting as I implement the loss function and dataloader to help finally get to training!

Code is here. See you for day 9!