unit1 ex1: don't skip special tokens

[belambert]

I think this flag might have been accidentally flipped from skip_special_tokens=False to skip_special_tokens=True at some point. It works much better with the subsequent line of code this way. I changed the code in both the notebook and the markdown, but I only updated the output in the markdown (not the notebook).

[Copilot]

Pull Request Overview

This PR fixes a bug where the skip_special_tokens parameter was incorrectly set to True instead of False when decoding tokenizer outputs. The change ensures that special tokens like <|im_start|> and <|im_end|> are preserved in the decoded text, which is necessary for the subsequent parsing logic that extracts assistant responses.

• Changed skip_special_tokens=True to skip_special_tokens=False in tokenizer decode calls
• Updated output examples in markdown to reflect the corrected behavior

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
units/en/unit1/4.md	Fixed tokenizer decode parameter and updated output examples to show correct behavior
notebooks/1/4.ipynb	Fixed tokenizer decode parameter in notebook code cells

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[belambert]

@burtenshaw