I think you can get what you want with BING Chat which is more advanced ChatGPT and can access internet.
1 Like
When they tell you that ChatGPT is not connected to the internet, other than by the means to communicate with users, they do mean it. It can only work with information that it was trained on, or that is in the prompts it receives.
If you give ChatGPT a URL, it looks at the tokens of that URL and attempts to predict what it thinks such a URL would be about, or what would be the most likely response, but it does NOT access the URL or use what is actually there, unless it was somehow already a part of its training data.
It’s incredibly clever, and it has fooled many into thinking that it is a work-around that allows ChatGPT to access URLs, but it really isn’t.
ChatGPT can process and summarize what is in a prompt, limited by the size of prompt data it can receive in any one ‘chunk’. By breaking down long texts into manageable chunks, you can summarize those piece by piece, but no, they are not currently able to allow millions of users to upload 150 page documents to process all at the same time.
5 Likes
So it means there is any way?
I think based on what you said which is true, many are deceived with the extension “WebChatGPT” that it makes ChatGPT understands and accesses URL topics resulted from the extenstion output, where it ofcourse can not do that.
What happens is the extenstion searches on Google with the input by user, and returns the summary of the URL which is also from Google 
.
1 Like
Start with a prompt like "Summarize the following text into no more than 600 words while retaining all key points: " followed by pasting one page of the pdf. Repeat for each page.
You may find that you want more or less than 600 words of summary, depending on the length and complexity of each page, but that should give you a starting point.
Personally, I’d suggest a secondary process where you then re-process 2 or 3 of the outputs at a time, further refining the outputs, and helping to smooth the ‘rough edges’ where one page may not alone have had all of the context necessary.
1 Like
this process needs more time of reading book
do you know any information about this php Code ?
1 Like
trying to find a way 
2 Likes
There are tons of language models you can run on your own machines or servers that allow you to process any amount of data your machine can physically handle. That’s how LLMs themselves are made, by running dozens, scores, or even hundreds of high-end computers over immense amounts of data to tokenize it all into a manageable system.
But ChatGPT isn’t running on your own servers or computers, and you are sharing it with millions of other users, all of whom have their own demands and needs. So you are limited in what it can process, so the whole system doesn’t crash.
3 Likes
may i ask you, advice me some of them ?
Hi, Im a bit of a newbie but could you take the PDF and copy the text (or convert it into a .DOC) and then put it into ChatGPT splitter As I understand, this splits the text into suitable chunks of text for ChatGPT However, if this a big docunent jt nay be unwieldy.
Quick question. If i do use ChatGPT Splitter and want to rewrie or condense akk the chunks of text, what is an effective prompt for doing so? Thanks.
1 Like
Hi, I used the splitter on a very large file. I see it segmented into 300 sections. So, I guess I would copy and paste as many of the segments that I want into chatgpt?
Have you encountered a prompt to tell it how to absorb the content? Something like, “I am going to paste in text by segments, one after another. I will tell you when I am finished”. Something like this?
1 Like
The problem with this wish: only 3000 words memory
2 Likes
There is a way out. Cut the PDF in 4000 tokens chunks, index the chunks by the embeddings, and look for the most relevant to your question. Use that chunk as prompt behind the scenes and answer. That’s what chatPDF does and here is the explanation.
1 Like
Welcome to the AIPRM community and thanks for your valuable contribution.
The mentioned GPTindex rebranded to Llama-Index must days before Facebook launching “their” Llama.
I watch it closely and cannot wait to get my feet wet with it. The success of AIPRM has pushed this back a little
2 Likes
I’ve also come across this issue but I solved it.
There are a few plugins on WordPress that take your data and send it over to Pinecone.io to have it indexed and available for search as an embedding. It also does this by sending over all your posts/pages to be embedded as well. So you can now run a chatbot off all your site’s data, or create a page with the content of your PDF, that gets sent over and now you can create a chatbot on your website. The plugin also reads the page content on the page that’s being loaded and uses that as context for the bot. If you have been using WordPress like me for years, this is pretty easy. Any questions, give me a shout.
Just tried ChatGPT Splitter, seems to work quite well from text, not so much from uploading a PDF.
3 Likes
1 Like
Thanks Stephen - I’ll try your suggestion out and experiment further. Cheers!
It does not work very good!!! The information is not verify. Sorry!!
You can try…
1 Like
I use this tool: https://chatgpt-prompt-splitter.jjdiaz.dev/, and it works just fine. It’s easy to use as it splits up the allowed text length into parts with just one copy-click.
"ChatGPT/AIPRM: Part 3/6 received.
Awaiting the next part.
_ _
Part 6/6 received.
Thank you for providing all the parts. I have received the complete text and can now continue processing your request. How can I assist you further with this information?
2 Likes