1739145600000

Work achieved

After starting to read the book previously mentioned, I understood that I could use Keris as wrapper for torch (because of its straightforward API) to chain an embedding layer with LSTM layers to train a models to predict the next characters for my sequences of characters. I asked an LLM to provide me with a prototype and started testing how the architecture behaved.

Discussions

Bill tested my knowledge of the code that was in the prototype. I was unable to explain what the shape of the X and Y of the model look like. As I was unable to answer those question it was decided to focus on understanding that prototype more in depth.