The amount of text you vectorize matters!

The vector we showed earlier came from the text What are the advantages of using RAG?. That is a relatively short amount of text, which means a 1,536-dimension vector is going to do a very thorough job representing the context within that text. But if we go back to the code, the content that we vectorize to represent our data comes from here:

loader = WebBaseLoader(
    web_paths=("https://kbourne.github.io/chapter1.html",),
    bs_kwargs=dict(
        parse_only=bs4.SoupStrainer(
            class_=("post-content", "post-title",
                    "post-header")
        )
    ),
)
docs = loader.load()

...