Happy almost Friday! It’s Day 18 of building my LLM from scratch ✨ So far we have seen the 1st stage of data preparation, sampling, coding attention mechanism and creating the GPT architecture. In the next few days I will be pre training my LLM. Today was all about Implementing the training loop, running initial evaluations, and watching untrained generations fail (as expected). This gives a baseline for tracking learning progress during pretraining. #deeplearning #generativeai #womenwhocode #largelanguagemodels #womenintech