The code in line 106 says - search_results (List[str]): Text content from web pages as search results.
But in line 123, it looks like a map soup = BeautifulSoup(html_text['page_result'], features="html.parser")
@jiazunchen : Thanks for pointing it out. We have fixed the doc strings to refer to the updated dataschema in this commit.
Please pull in the latest changes from the upstream starter kit.