@vincerocchi9083
Appreciate the effort you put in in putting this together! I recently completed my first vector rag pilot and I’m looking to expand into a graph rag option. This was great background with plenty of detail and thoughts on the different levers a designer can pull to get different types of results. Excellent job and thank you!!
@dtesta
Wow, your voice is made for explaining. Very clear and melodic. Well done!
@KrisGlowacki
Amazingly useful content as usual. One of the best channels for AI devs. Good job Adam!
@PauloHenriquedaSilvaMota
I am doing my thesis about knowledge Graph and this video was a great start in the topic!
@jimlynch9390
That was very interesting. It's an approach to searches I have never thought about. I'll probably not use the info directly but having knowledge of it's existence may be valuable one day. Thanks.
@Alistair
This is some super high quality content, and covers just what I was wondering about recently. Thanks!
@lavamonkeymc
Love the graph rag vid, probably one of the best channels out there now, keep up the interesting content like this, some other examples you can do is fine tuning LLMs with llama factory (if you haven’t done that.). Also, LiteGraphRag just came out from MSFT which optimizes the indexing and retrieval for less than 10% accuracy. One thing I struggle with is , we do graph rag to get relationships or summaries between data, but we can technically do this on a rag-level too, by summarizing large chunks of rag content and storing as metadata or another chunk to use during inference. Graph rag will always have better structural relationships, but advanced rag appending/metadata summaries of content will be a close second with higher-level questions like these.
@vn_vn.vn-vn_vn
Amazing content, brother! Simple, useful, and well explained. Keep up the good work!
@thetruenosato
This is my type of content, direct and clear. Thank youu!
@leugim8872
You sir a gentleman and a scholar, thank you for the overview.
@jonathan_blackburn
Cool. I love that clean code and the explainable illustrations!
@EliSpizzichino
Great video! People with some years like me, see this strongly similar to Resource Description Framework (RDF) and related technologies
@JurassicMonkey-x9f
Adam, another great presentation! Howsabout some hard numbers comparing information retrieval scores for graph rag vs some other rags? Cheers!
@AdamBrusselback
Thanks for the video. I did some work on graph generation like a year and a half ago before any of the solutions came out. Had a solid solution to graph creation from single chunked documents of any length, but never got around to the multi document merging step. Glad to see how far these solutions have come, and that I stopped working on something that was going to be eclipsed right away. Though I was more focused on reasoning using the created graph rather than using it for RAG. I should dig out that dataset and train something more modern than llama 2 13b.
@kargarisaac
Amazing. Any plan to review LightRAG as well?
@aymasrayman1190
Please what softwar or package did you use for the visualization of the graphs ? Great video as usual !
@vivi6432
Insightful & Comprehensive!
@LibraryofCelsus
Hate how Microsoft claimed they came up with these concepts. Both have been around in more simple forms for over 2 years in the open source community. Really shows that in terms of knowledge and skill, they are not as ahead as everyone thinks. They just have more resources.
@k9clubme
@Adam thank you so much for sharing your expertise. Love the videos and I learned so much.
@nexuslux