Can your app be used to import 1k blog posts, categorize them and show a visual graph some sorts of? Maybe extend them via ChatGPT aided research questions and summaries?
Yes, sure, you can upload a set of documents and find the gaps between the ideas within. You even have a view where you show the documents as the nodes in the graph, but in your case I don't think it's necessary.
What's the maximum size PDF that can be uploaded? I would like to look at my own work from this perspective, however, the document in question is ~250pages...
I think it can handle it, but if it says it's too big, simply convert it into a text file and reupload as a .txt. Usually a book of 250 pages is less than 1 Mb, so it should work!
Why not, yes. In fact, the approach itself doesn't need any LLMs. You just upload a doc to InfraNodus and then look for the gaps in the structure. It tells you what the topical clusters are, so you can then generate ideas based on these insights. You only need an LLM to come up with an idea for you. It's like mental prosthetic, but you don't really need it.
Ok, but uploading my obsidian vault is a no go. That's why people want to use everything locally because they don't want to share their private data anywhere. Can one use Your tool completely offline ad local standalone app?
@@TheMastish Exactly.... I consider my entire vault one "doc", and I am definitely not uploading 2 million + words....The whole vault is geared towards one project for which I wish to communicate holistically with, not individual uploads at a time. If you or anyone else here is in Los Angeles by any chance I would love to converse over such topics in person.
@@TheMastish Would you consider a system similar to LogSeq, where you can give access to your folders in a browser to analyze the files, if nothing is stored on the cloud?
how does this work in the sense of querying not only within the document but also the gpt model on its own? if i understand it correctly, usually its either looking into the document, or not at all. can we do both at the same time? /edit: ah, 5 seconds later you explained the check box.. could you imagine to implement a parallel process? like querying the document, querying the model on its own, then have the model combine both outputs.
Hey in fact at the moment it’s realized as you are saying: you query the outside context if the checkbox is off and then if it’s on then you take the document into account and kind of combine the both contexts (internal and external). Is this what you had in mind?
@@noduslabs yes exactly. i was asking because in the demo it seemed as it was exclusively reffering to the context of the document. and in other applications thats usually how its implemented, right, there is a prompt which tells the model to either reply within the context of the document or to not take it into account. i thought it would be interesting to do this in parallel and then have to model combine that and output something more complete, taking a wider context, outside the documents subjects, into account. does that make sense? by the way, great work, still need to figure out how to implement this into my process, but its a powerful and quite beautiful tool!
Can your app be used to import 1k blog posts, categorize them and show a visual graph some sorts of? Maybe extend them via ChatGPT aided research questions and summaries?
Can the tool interact with a set of documents rather than just a single doc? So I can find gaps across all of my documents?
Yes, sure, you can upload a set of documents and find the gaps between the ideas within. You even have a view where you show the documents as the nodes in the graph, but in your case I don't think it's necessary.
Neat - CAN you upload several PDFs? different documents? 🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌🙌
Yes, it's all possible!
How is the knowledge graph built?
You can find the complete explanation of how it works on our page infranodus.com/about/how-it-works
Fantastic
amazing
What's the maximum size PDF that can be uploaded? I would like to look at my own work from this perspective, however, the document in question is ~250pages...
I think it can handle it, but if it says it's too big, simply convert it into a text file and reupload as a .txt. Usually a book of 250 pages is less than 1 Mb, so it should work!
Homicide Detectives have been looking for something like this for a long long time
Are you one of them? I'd be curious to explore a real use case.
Can i insert like multiple pdfs and analyze the common content for example? Is that possible?
Sure, it's possible. You can upload multiple files, PDFs, txt, MDs, etc.
Bro this is awesome. Really thoughtful.
Thank you! Nice to hear it!
Can this be done with offline LLM'S?
Why not, yes. In fact, the approach itself doesn't need any LLMs. You just upload a doc to InfraNodus and then look for the gaps in the structure. It tells you what the topical clusters are, so you can then generate ideas based on these insights. You only need an LLM to come up with an idea for you. It's like mental prosthetic, but you don't really need it.
Ok, but uploading my obsidian vault is a no go. That's why people want to use everything locally because they don't want to share their private data anywhere. Can one use Your tool completely offline ad local standalone app?
@@TheMastish Exactly.... I consider my entire vault one "doc", and I am definitely not uploading 2 million + words....The whole vault is geared towards one project for which I wish to communicate holistically with, not individual uploads at a time.
If you or anyone else here is in Los Angeles by any chance I would love to converse over such topics in person.
@@TheMastish Would you consider a system similar to LogSeq, where you can give access to your folders in a browser to analyze the files, if nothing is stored on the cloud?
how does this work in the sense of querying not only within the document but also the gpt model on its own? if i understand it correctly, usually its either looking into the document, or not at all. can we do both at the same time? /edit: ah, 5 seconds later you explained the check box.. could you imagine to implement a parallel process? like querying the document, querying the model on its own, then have the model combine both outputs.
Hey in fact at the moment it’s realized as you are saying: you query the outside context if the checkbox is off and then if it’s on then you take the document into account and kind of combine the both contexts (internal and external). Is this what you had in mind?
@@noduslabs yes exactly. i was asking because in the demo it seemed as it was exclusively reffering to the context of the document. and in other applications thats usually how its implemented, right, there is a prompt which tells the model to either reply within the context of the document or to not take it into account. i thought it would be interesting to do this in parallel and then have to model combine that and output something more complete, taking a wider context, outside the documents subjects, into account. does that make sense? by the way, great work, still need to figure out how to implement this into my process, but its a powerful and quite beautiful tool!