KeyError: 'content' about gpt-author HOT 2 CLOSED

This is (probably) a rate limit error. The notebook as provided has absolutely no error reporting/handling, which makes it difficult to be sure, but it's the most likely culprit. I would suggest, if you want to play around this and don't want to give Anthropic lots of money to increase your rate limit,

• request a smaller number of chapters
• in the line chapter = generate_text(chapter_prompt, max_tokens=4000), lower the tokens to 3000

If you want to delve into Python a bit more, I would also suggest adding some basic error handling to generate_text, and maybe retry on a 429 or 529 error (which are hitting your rate limit and Anthropic's API being overloaded, respectively). I used something like

if response.status_code != 200:
    print(f" ! got status code {response.status_code}")
    print(response.json())
else:
    response_text = response.json()['content'][0]['text']
    return response_text.strip()

…but I also broke the notebook down into a series of cells rather than just one big script, because I wanted to be able to investigate just what's being generated along the way to see things like "how good is the outline Claude-Author has generated from the prompt" and "how well is Claude-Author following its own generated outline as it creates the chapters".

from gpt-author.

Thank you very much for your reply. I reduced the number of chapters to 10 and now the books are generated without errors)

from gpt-author.