I've been trying to get ChatGPT to pull quotes from various books, but it keeps making up sentences that aren't actually in the texts. I'm not looking for it to summarize; I just want it to analyze the content accurately. Anyone else experienced this issue, or have any tips for better results?
4 Answers
The issue largely comes down to ChatGPT's context window limitations. If you're only using the free version, you're stuck with lower token limits which make it tough to fit everything when analyzing complex texts. Smaller chunks or switching to models that support more context might help.
You might want to try using different models like Gemini or NotebookLM. They tend to handle larger texts better and have longer context limits, which can help with accuracy in analysis. ChatGPT has some limits on what it can remember from longer works. Shift to those other tools for better results!
Thanks for the recommendation! I’ll check those out!
Heard that the fewer tokens you feed into ChatGPT at once, the better the results can be, especially for detailed analysis. If your works exceed the standard limits significantly, just chunk them down into manageable pieces for upload. Gives much clearer responses!
Thanks for the tip! I didn’t realize token limits were affecting things so much.
It sounds like ChatGPT might not be recognizing the full context of the book. It typically relies on summaries rather than remembering text verbatim. For more accurate analyses, consider uploading chapters one at a time instead. That way, you can maintain context while asking your questions.
Great idea! I’ll try feeding it chapter by chapter.
Totally agree! It’s smarter to break it down into smaller sections, especially if the books are lengthy.