Query Your Own Notes With LangChain

Key Takeaways

what's up guys in this video we're going to learn how to use Link Chang to ask questions to your own notes let's get started first thing that I'm going to do here is I'm using obsidian as my note-taking app so I'm going to be loading the obsidian loader from linkchang.document loaders then I'm gonna set up the path to the notes where I have everything stored so in this case it's this path where I have my notes that I'm showing here on the right and now I'm gonna load those notes into a link chain using the loader variable that I initialize with the obsidian loader class now I'm going to create my index and right now according to link chain the best best index you can have is Vector store index so that requires chroma DB which you should have installed in your computer you can just run in a notebook pip in stock from a DB then I'm going to load the vector store index Creator and then I'm gonna access The Dot from loaders method giving the loader variable that I initialized here with the loader and create my index perfect now that I created my index the fun part starts I mean it's amazing how easy and simple it is to to do this on your computer so this is an approach that I think is going to change the way that we take notes and that we think in terms of like how we reflect with the computer because now I can just ask a question like what are the a problems I've listed on my 12 problems note about AI research because if I come here to my notes I have a note called 12 problems that I was trying to you know I remember there was this thing about Feynman and a technique that he used to use which was to describe all the 12 main problems that you thought about that you reflected on and I thought it was cool and I started making this list so it has like problems like intelligence generalization with his attention probably credible assignment these are classic problems in artificial intelligence research that I'm interested in that I think are fascinating so I'm asking something very specific what are the problems that I have on this node so now I can say this and now I can use the query with sources method which is gonna get my question and then send it to the index that I created before and return the result so now when I run this there we go so this is the question and the answer is the a problem is listed on the 12 problems node are problem intelligence generalization problem what is attention credit it's it's perfect and it's getting exactly what I needed which was these notes so it's just amazing how it works considering that I'm asking something with natural language right I'm not doing a control F search on my notes I'm asking something to my notes as if my notes you know were in as if I was talking to someone they've read all my notes and remember all my notes and can actually tell me something about what I've written before so this is really cool so now let's let's just create a few more queries so that you guys get a sense for how amazing and Powerful this approach can be for note-taking for productivity for all sorts of stuff so in this one I'm saying summarize my thoughts on technology because I created a note here called thoughts on technology yeah this was just like some random few paragraphs that I just created with judge PT just to see what happens when we ask a question like this so let's take a look technology can be used to augment our capabilities allowing us to represent complex thoughts and pieces of information into Outsource mental capabilities to reduce the effort into Discovery it's important to approach the advancements on the balance of enthusiasms and caution ensuring that co-conspirations and societal implications are addressed so this is really cool because it's getting essentially it's coming here to this node as well as this note called path to focus which I didn't even remember that I had oh this is so amazing and it's combining them meaning it's giving me insight about what I've written that I don't even remember that I wrote and it's really cool because I actually just tested this this was the first time that I tested this this particular query and I did it based on the thoughts on technology nodes so the fact that it kind of like direct me to the to this note that I wrote that I've written a while a while ago is just amazing it's really cool so now let's do another one so this is a bit more precise so this one was more reflexive now let's do one that's more precise so I'm saying give me the main points of my note about automating data science workflows because I had this idea that you know with length chain and these automations are coming up using large language models you know one of the things that's really interesting is how you can automate data science workflows using these kinds of tools so I started writing a few ideas on this topic and I'll show you the note here on my right so this is the note it's a big node I generated with judge PT and then I you know did some magic on it so you have these components problem formulation data set understanding data preprocessing feature engineering Etc right so I'm trying to retrieve those specifically from this node uh so let's see what happens so now I'm querying and let's see the main points of the note about automation Mark flows are leveraging natural language processing and AI driven tools to streamline automate various aspects of data science workflows okay all right so it's kind of like it understood that it's the part about understanding I wish it would have said just problem formulation data set understanding but okay and then using a power language models like jgpt to assist data settings that's like data exploration uh um okay future engineering uh let's open up this thing so that we see it a little bit better section and Report generation okay so it got this part perfect model selection it got this part perfect uh uh model interpretation evaluation let's see okay so the this summary with the points wasn't as perfect as I was you know assuming just because it wasn't comprehensive but it was it was pretty good it was pretty good it got everything almost perfectly right so yeah that's it that's pretty much it this is the power of using link chain for your notes you can essentially query your notes using natural language which for me I think it allows you to have like some Next Level insight into your own thoughts because you can start writing of course you can always connect your notes physically but join these connections by having some kind of model with whom you can chat is really impressive so if you like this video don't forget to like And subscribe and see you next time cheers