I think for non programmers like me... this might still be too next level XD
@cgc230018 күн бұрын
hello I wrote to you to tell you that I have a lot of difficulty understanding how GPT works
@cgc2300Ай бұрын
Hello
@cgc2300Ай бұрын
Hello Could you give me examples so that I understand how the APIs work or more precisely the results that I could obtain depending on the different apps that I use? Do you know if there is a site that can give me examples of using GPTs with API just to have examples?
@n-2094Ай бұрын
Thank you for the great video!💡🙌
@cgc2300Ай бұрын
hello is there a GPT to which I could give my ideas and my objectives and that it tells me how to build a m on where the prompts and if I should add a database and APIs? I ask the cat but he doesn't know how to do it, I don't think he's very intelligent and thank you for the videos
@cgc23002 ай бұрын
hello and thank you very much for your videos, however I still have a lot of things that I don't understand, could I ask you some questions
@hubel-labs2 ай бұрын
Feel free to ask here and I will try to answer if I can
@cgc23002 ай бұрын
@@hubel-labs I have around ten ideas and the first concerns SEO and automation. I don't know how to go about it and I would like to know if my ideas are good and feasible. I talk about my ideas to chatgpt so that he can tell me what he thinks about them and then correct me if these ideas are not good or so that he can make suggestions to me and also he can tell me how to make a GPT based on my ideas and my goals but these answers are very basic I would like to sell on eBay... my first idea I had regarding automation an idea would be that my analytical software scans all my listings on eBay then automatically scans me and searches for keywords then changes all the keywords automatically without my intervention so it's an idea that I had and I submitted it to gpt4 so that he can tell me if it's a good idea and if there is perhaps a better one to offer me but these answers are are very basic so can you tell me what you think about it? This is only my first idea but I also had many others and I wrote them on a Google doc but the problem is that I asked chatgpt to translate it for me into English understandable but he is also very bad at it I mean the translations remain very basic like with Google Translate
@cgc23002 ай бұрын
@@hubel-labs I have around ten ideas and the first concerns SEO and automation. I don't know how to go about it and I would like to know if my ideas are good and feasible. I talk about my ideas to chatgpt so that he can tell me what he thinks about them and then correct me if these ideas are not good or so that he can make suggestions to me and also he can tell me how to make a GPT based on my ideas and my goals but these answers are very basic I would like to sell on eBay... my first idea I had regarding automation an idea would be that my analytical software scans all my listings on eBay then automatically scans me and searches for keywords then changes all the keywords automatically without my intervention so it's an idea that I had and I submitted it to gpt4 so that he can tell me if it's a good idea and if there is perhaps a better one to offer me but these answers are are very basic so can you tell me what you think about it? This is only my first idea but I also had many others and I wrote them on a Google doc but the problem is that I asked chatgpt to translate it for me into English understandable but he is also very bad at it I mean the translations remain very basic like with Google Translate
@cgc23002 ай бұрын
@hubel-labs it's very complicated to understand how actions and APIs work I have around ten ideas and the first concerns SEO and automation. I don't know how to go about it and I would like to know if my ideas are good and feasible. I talk about my ideas to chatgpt so that he can tell me what he thinks about them and then correct me if these ideas are not good or so that he can make suggestions to me and also he can tell me how to make a GPT based on my ideas and my goals but these answers are very basic I would like to sell on eBay... my first idea I had regarding automation an idea would be that my analytical software scans all my listings on eBay then automatically scans me and searches for keywords then changes all the keywords automatically without my intervention so it's an idea that I had and I submitted it to gpt4 so that he can tell me if it's a good idea and if there is perhaps a better one to offer me but these answers are are very basic so can you tell me what you think about it? This is only my first idea but I also had many others and I wrote them on a Google doc but the problem is that I asked chatgpt to translate it for me into English understandable but he is also very bad at it I mean the translations remain very basic like with Google Translate
@cgc2300Ай бұрын
to sum up : there are several topics that I would like to look into and I have 1000 questions, I would like to use gpt4 like a super teacher, to have him correct my ideas about marketing, and it's really not great for me despite the 100 billion pieces of data it has and I understood that if I do a gpts then the problem will be solved? there are directories of GPTs but what I would like is to see how they are made up for example there is a GPT it is called Universal primer and which apparently is the kind of GPT I need I mean who knows everything about everything and who can give me amazing tutorials, you know? but there is no way of knowing how this kind of gpt is done? also I would have liked to ask GPT 4 how to make a GPT according to my objective but gpt4 and as evasive as 3.5 and still in the construction of a GPT, I would have liked to know what it is possible to do with the APIs So what I would like is to tell you what I want to do, my objectives, and for you to tell me how I should build my GPS or is there a gpt that could help me say how Do I have to create my GPT for what I want, I fart 4 can't do it despite its 100 billion data
@AnoniChocolateMoose2 ай бұрын
Thankyou
@victorfeight96442 ай бұрын
clear effective explanations thank you
@karanv2934 ай бұрын
I have a question because theres so many tutorials out there regarding "Advanced RAG strategies". Im trying to find the best one for my usecase which is users ask questions over 100s of large documents. Theres a tutorial on llamaindex youtube called Llamaindex Workship: Multimodal + advanced RAG workshop with gemini. how does this compare with that strategy?
@user-tc2ef9ek5r4 ай бұрын
Great video!
@NetZeroEarth4 ай бұрын
Hi Lily How do I gain access to the colab book. I'm following Huber labs on LinkedIn under Jamall Winn
@SwingingInTheHood4 ай бұрын
Thanks for the video. It helped me get my GPT and action up and running within a couple hours. Good work!
@Ben-gp5ty4 ай бұрын
Can we use AWS sagemaker jumpstart to serve the E5 embedding model ? Are there advantages of doing it ourselves instead like what you are doing ?
@ginisksam5 ай бұрын
Llama Index has new version 0.10 - will migrate your codes n learn same time. Thanks for introducing Sentence Window Retrieval. Most basic straight-split and retrieve/chat doesnt produce much meanings on our docs.
@JOHNSMITH-ve3rq5 ай бұрын
Ah loving this chan. Forgot about it. It’s the bomb!!!
@sivi38836 ай бұрын
Awesome video explained very clearly! Thanks a ton! If I may ask, what tool do you use for those visual flows. Love it!
@sayanbhattacharyya956 ай бұрын
Really cool video! Is there an "ideal" or "recommended" value of window_size?
@jannessantoso14056 ай бұрын
it is 1784 in teahistory.txt and 1794 in chinahistory.txt so bit confusing but anw great tutorial Thanks
@fheralanis6 ай бұрын
your video has helped me too much to find the source of the messages that the assistant generates when the 'run' is 'in-progress' using the 'code interpreter'. thank you very much for sharing !!
@kashishvarshney22256 ай бұрын
amazing video mam, please tell me one thing, I created a chatbot using this method but it takes almost 15-20 second to response. Can I decrease this response time please reply
@nazihfattal9746 ай бұрын
As always, well prepared, easy to follow video that delivers a lot of information and value. Thank you!
@AssassinUK6 ай бұрын
Ive been doing this but using FlowiseAI
@thisiscrispin6 ай бұрын
You mention providing examples in the end, but do I see that correctly that you never feed those into an API call to openai?
@hubel-labs6 ай бұрын
U mean the few shot learning part? At 11:33, and also at 13:36, when I ran those cells, the examples did get passed in to the API call to get the json output result. Not sure if that helps to answer your question. The way that I’m using function calling here is a bit of a hack. If you want to see how it is intended to work, can watch this one: OpenAI Assistants API - Part 2: using Agents with function calling together with info retrieval kzfaq.info/get/bejne/eLxdkpiQ15qam3k.html
@raveeshnilaweera6 ай бұрын
Thank you ❤
@WesleyMusgrove6 ай бұрын
Hi this was super helpful and I'm looking for you on LinkedIn, but there are a ton of Lily Cheng's, so I'm not sure which is you. Would love to get a copy of this code to learn more!
@goonymiami6 ай бұрын
Seeing your explaination at around 09:30, it seems like we can only use K windows to serve as knowledge base to answer a prompt. What If the prompt asks information that is contained in more than K windows? Like if I have several documents containing each a bio of a person, and if the user asks to sort those 10 people by age... how can it figure it out? I guess we can use a big value for K, if the cosine similarity engine can take it... but I am guessing providing too much context to the LLM will cost a lot of money?
@hubel-labs6 ай бұрын
Yeah, it wouldn’t work very well in that scenario. I wonder if perhaps there is a strategy where the LLM can determine the right K based on what it is trying to do. Or instead of setting K to be a fixed number, to return all chunks where the cosine similarity is above a given threshold.
@mikestaub6 ай бұрын
Incredibly useful, thank you!
@JOHNSMITH-ve3rq6 ай бұрын
Text very distracting !! Killer content otherwise tho.
@JOHNSMITH-ve3rq6 ай бұрын
Astonished by the high information density in this video. Needs 10x the views. Amazing. Thank you.
@nunoalexandre64087 ай бұрын
Love it!!!!!!!!!!!!!!!!!
@mikestaub7 ай бұрын
This is currently the best RAG tutorial on the internet.
@Work_Pavan-mu9ye7 ай бұрын
Excellent video. Liked the workflow you showed in the beginning. What SW are you using to create this workflow?
@hubel-labs7 ай бұрын
Excalidraw
@unclecode7 ай бұрын
Fascinating! Your approach to teaching and presenting is poetic. It is well organized, well explained, and well illustrated. Indeed, kudos to you. If I could, I would subscribe to your channel twice!
@JOHNSMITH-ve3rq7 ай бұрын
Wow _ darn useful !!
@chiggly0077 ай бұрын
Great video? Is the diagram anywhere to refeeence?
@hubel-labs7 ай бұрын
I’ll post it up tomorrow - need to fix an error
@grabani7 ай бұрын
@hubel-labs Any update on the diagram please - thanks? Great vid by the way.
@hubel-labs7 ай бұрын
I added it to the description. Here is the link: link.excalidraw.com/readonly/m6DK7oyEFpyQnuw55DVP?darkMode=true
@natal107 ай бұрын
Great video!! Love it
@hubel-labs7 ай бұрын
Thank you!!
@daaimrehman7177 ай бұрын
can you people please make a tutorial of a project using gemini? like a simple financial analyst or something?
@danielvalentine1327 ай бұрын
Great job explaining window and how vector store and doc store relate and where window lives. I’ve been trying to understand this aspect of llamaindex, and you made it very clear!
@hubel-labs7 ай бұрын
Yeah, it took me a while to figure that out too. Glad it helped you!
@user-jj1ff2gz7u7 ай бұрын
df
@sitedev7 ай бұрын
Great video with an awesome easy to follow explanation of RAG. Reminds of a recent Andrej Karpathy video.
@eternalia40597 ай бұрын
Thanks, great job
@peteredmonds17127 ай бұрын
I'm a bit confused on the use case for re-ranking. Doesn't that defeat the purpose the top-k search in that we include all chunks, significantly increasing the number of tokens we use? Is the idea to do re-ranking with a smaller & cheaper LLM before sending the resultant top-K chunks to a more robust LLM?
@hubel-labs7 ай бұрын
yeah, it took me a while to understand it as well. It does end up using more tokens but I guess that can be, as you said, mitigated with a lower spec LLM. The idea is that embeddings is not a perfect representation of the text - it's a good way to sort through millions of text chunks efficiently but the exact ranking within the top 20, for instance, may not be as good as using the text directly. So if you were to do reranking, you might use the embeddings to return j results (where j is somewhat larger than k) and then rerank those j results using LLM into a final top k that you then pass into your user-facing LLM.
@JamesRBentley7 ай бұрын
Great demo, and Lily is awesome, I dropped her a line on LinkedIn and she kindly shared the colab workbooks.
@yourbestsail7 ай бұрын
Absolutely excellent! Finally a crystal clear explanation! Thank you 😊
@peteredmonds17127 ай бұрын
Small correction: embedding are not a 1536 digit number but of vector of size 1536
@hubel-labs7 ай бұрын
Yes, you are right …. I was thinking of hashes for some reason! I’ll add a correction!
@MaliRasko7 ай бұрын
Your explanations and delivery is on point. Thank you for an excellent content and relaxed narration style.
@mayurmoudhgalya38407 ай бұрын
Another great video!
@arjoai7 ай бұрын
Wow, that’s a wonderful piece of advice from such a talented professional in the field. Thank you 😊
@Scorpio2011b7 ай бұрын
Hi, thank you for the very informative vidoes i honestly saved alot of time, and the diagram reslly simplified how to understand it all. I've just sent you a request on LinkedIn but can't message you for some reason.