Minor Lab issues in Week 2

Course Q&A Generative AI with Large Language Models
1

Hi!

Two things I noticed that might be worth looking at.

  1. In the various training steps…
    num_train_epochs=1,
    …
    max_steps=1

are both specified. I was looking up what these parameters do, and according to the Hugging face docs, if max_steps is specified, it overrides num_train_epochs.

  1. There is an error in the notebook at 3.5 Evaluate the Model Qualitatively (Human Evaluation).

    The variable is assigned as baseline_human_summary at the top, but what is later evaluated is human_baseline_summary. Darn global variables, eh?

2

Hi Justin, and welcome to the community! Thank you for the feedback. The team will review and make revisions accordingly.