jupediaz / chatgpt-prompt-splitter Goto Github PK

Though you can indeed split up a long text into chunks, and you then feed them all into a ChatGPT chat, this is irrelevant.

The maximum input length in a ChatGPT prompt is equivalent to the maximum context size for ChatGPT. Thus, ChatGPT will "forget" all the previous snippets, and it will only know about the final snippet.

Try it: use this script to break apart a long text. Then quiz it on a detail only present in the first snippet. ChatGPT will have no idea about it, since it's outside of its current context, which is ONLY THE FINAL SNIPPET. (And maybe a bit of the second-to-last snippet, if the final snippet is less than the maximum input length.)

This problem is impossible to solve in a fully satisfactory way. I'd love to have a longer context, too, but this is unfortunately not solvable by breaking apart a long text. The best you can do is have ChatGPT iteratively summarize previous snippets, then feed the summary into the next snippet. You can programmatically do this, and it might work for your purposes.

As for this project, it should have a disclaimer at the top that this script is useless / just for fun, because it's currently just wasting people's time. If you want to salvage this project into something useful, then here's what it could look like:

• Allow user to input the full text of their long document.
• Output the first part of the document, along with a request to summarize.
• Allow the user to input the summary of the first part.
• Output the next part of the document, along with the summary, and request ChatGPT to summarize both.
• Repeat until the entire long document is summarized.

The above is a script that would indeed be useful. But the current one is a waste of time and should ethically have a disclaimer at the top.

Thanks for your great idea.
I tried with calling OpenAI API and sent chunks into the conversation context.
Due to the token limit, I got this error:

"type": "InvalidRequestError",
"message": "This model's maximum context length is 4097 tokens. However, your messages resulted in 4415 tokens. Please reduce the length of the messages."

Any suggestion?

Are there any plans to find a way to use this with the openai api, rather than the chatgpt frontend?

I don't understand how do you guys run this? I get all sorts of errors, from .env (not included) not found to index.html template not found (wrong folder?).

Hey, can you make a script that Auto paste the chuncks to chatGPT??

Great script, really appreciate your work!

Issue Description:

The "Split into" feature in the demo at https://chatgpt-prompt-splitter.jjdiaz.dev is incorrectly calculating the number of chunks for large inputs. When pasting text with approximately 10,000 characters, the application suggests splitting the text into 11 parts instead of the expected number.

Steps to Reproduce:

Visit https://chatgpt-prompt-splitter.jjdiaz.dev
Copy and paste text with around 10,000 characters into the "Enter the long prompt to be splitted" textarea.
Observe the suggested number of parts in the "Split into" feature.

Temporary Fix:

Select "Max chars length for each splitted part", switch to the "Custom" option, and then switch back to the default option.
Add an empty space into the "Enter the long prompt to be splitted" textarea.

Screenshots:

hey I love your tool, there is a thing, the past tense of split is split.

https://github.com/search?q=repo%3Ajupediaz%2Fchatgpt-prompt-splitter%20splitted&type=code

so maybe fix it, if you feel like it.

Will LangChain be supported in the future?

Clicking the button and using the standard Vercel prompting results in:

This Serverless Function has crashed.

Your connection is working correctly.

Vercel is working correctly.

500: INTERNAL_SERVER_ERROR
Code: FUNCTION_INVOCATION_FAILED
ID: pdx1::zf5z4-1685748399799-42778bd3fd88

Can this be ported to work offline on clientside?

Can't seme to get this to work properly via the deployed instance. Tried using GPT-3.5 and 4 but both still say the message submitted was too long.

15000 chars ~= 5200 tokens
Well after a few chunks, the initial text wouldn't be context anymore, right?

I've noticed the Instructions section always renders the page count as "1/10". Here's what is provided:

The total length of the content that I want to send you is too large to send in only one piece.
For sending you that content, I will follow this rule:
[START PART 1/10]
this is the content of the part 1 out of 10 in total
[END PART 1/10]
Then you just answer: "Received part 1/10"
And when I tell you "ALL PARTS SENT", then you can continue processing the data and answering my requests.

PS: I did read the "contributing guidelines" but I see no option to label this as a "Bug".