Great video! I would definitely be interested in seeing some evaluation methods. Cheers!
@khaledsaud6677
3 ай бұрын
This is great. Well organized with a clear explanation. I am lucky to find this channel. Keep up the nice work. I have one interesting idea I like to see how you approach it. Multi-agent framework with RAG to chat and perform some tasks with long PDF files such as legal documents. Specifically, I would like to see your chunking strategy (e.g. semantic chunking, agentic chunking, etc.) in this case. Tools can include document reviewer, researcher, compliance checker, etc. Also, please consider the Command R+ model by Cohere as well. It performs competitively with GPT4-turbo and offers up to a 57% cost reduction.
@mihaitanita
3 ай бұрын
Nice job! Two notes: - if this development continues, consider to calculate also the token consumption into the graph - in my scripts i'm "avoiding" Groq RPM by adding a sleep/wait param. (Groq is not the only service with this kind of limitation).
@jarad4621
3 ай бұрын
I use openrouter now, (groq rate limit sucks) grounded to the together ai provider using lllama70b and it's cheap as hell.and blaZingly fast and basically no rate limit, done very cool things with it. It's the best there is now. there is another service for llama 3 70b, $20 pm unlimited tokens you can run as many as you want perfect for testing hungry agents safely and cheap, let me know I'll try find the bookmark
@Data-Centric
3 ай бұрын
Great suggestion! Thank you
@kiiikoooPT
3 ай бұрын
yep, I was going to comment that second note, I never worked with groq but from what I seing from tutorials with CrewAI and groq, people use to add timers to prevent the rate limit.
@VinCarbone
3 ай бұрын
Absolutely the best Langraph example project explained on youtube!
@sirishkumar-m5z
Ай бұрын
The future of AI and technology can be daunting, but with the right tools like SmythOS, you can navigate these changes with confidence and stay ahead of the curve
@anmolmittal9
26 күн бұрын
Great Video!! I wanted to check how can we show these tool calls and other things as Chainlit steps?
@kalokali7711
3 ай бұрын
Very nice improvement !! Just wondering if there will be an option in the future to decide what model is assigned to specific agent - what I mean is, it will be nice to specialize agents with different models as of purpose of use eg. gpt4o || Gemini to get image details -- > claude to give nice summary and make a search of similar product in the net --- > local model (*ollama hosted) to make update on my local scripts for making newsletters. 🕸🏴☠
@CK.23.
3 ай бұрын
Great work. Thanx. However, is there a reason not to check if there is already key in config.yaml, and use it, instead of reset it every time?
@tee_iam78
2 ай бұрын
Great content. I have learnt a lot from the series of videos you have made throughout. Big Thumb Up.
@fraririri
3 ай бұрын
I've just started watching your videos, and they are amazing! Thank you so much for being so good at explaining difficult things. Keep it up!
@nanubalakundan9825
Ай бұрын
These vids are truly underrated!
@AbhishekSingh-pj1oo
3 ай бұрын
THIS IS GEM!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
@amrit448
3 ай бұрын
after selecting the model, and entering api key in chainlit, getting the error- [Errno 2] No such file or directory: 'G:/My Drive/Data-Centric Solutions/07. Digital Content/LangGraph/code/graph_websearch_agent/config/config.yaml' What did i do wrong?
@tee_iam78
2 ай бұрын
Thanks!
@Data-Centric
2 ай бұрын
I'm humbled, thank you!
@premusic242
3 ай бұрын
This is what I was looking for extremely helpful. Tired using streamlit from last few days but failed while using langraph
@jarad4621
3 ай бұрын
This is exactly what I need thanks a million
@realCleanK
Ай бұрын
Thank you!
@airobsmith
3 ай бұрын
This is really useful, but I feel like I am missing tool selection? The process is simple in this example, one agent decides on inputs for one tool, tool output goes to next agent, etc. This seems overly simplistic. If an agent might need to select from 2 or 3 tools would I need the agent to always talk to a router to select the correct tool? And then if the tool output might then require a selection of agent to process the result? Perhaps this example is better considered as a single "task agent" which is part of a larger agentic workflow?
@joaoloureiro9767
3 ай бұрын
Hey John, thank you for this awesome video/resource! I have a question for us that work with a dB aswell. How would you structure the file system and how would you implement handling dB requests and writes, for example, if you want to store the planner's agent thoughts in a dB, how would you achieve that with your modular approach? Thank you so much
@gileneusz
3 ай бұрын
have you tested Qwen2 for longer context capabilities?
@KingHack-vl7rj
3 ай бұрын
Keep up the nice work. Also do something with that attach/upload button in chainlit UI.
@guilhermeapolinario7860
3 ай бұрын
You have incredible teaching skills! Thank you very much for sharing, Good luck to you!
@IkechiGriffith
3 ай бұрын
Nice. This is dope. Love the changes and the rationale!
@Data-Centric
3 ай бұрын
Thanks! I saw your PR. Had these changes coming so couldn't merge. Thanks contributing though. Let me know if you have any success with Groq.
@IkechiGriffith
3 ай бұрын
@@Data-Centric Yea. No worries. I had a few challenges with this build. I’m on a Mac by the way so I’ll see what I can do to address some of them.
@deanchanter217
3 ай бұрын
What if you broke the reviewer down to it's individual tasks?
@pedrorafaelnunes
3 ай бұрын
Thanks, you ROCK !
@ManjaroBlack
3 ай бұрын
Nice. I found some of those issues. I’ll add SearXNG to this one as well 😂
@Data-Centric
3 ай бұрын
Nice one!
@morrazzo4432
3 ай бұрын
Before learning langgraph suggest me what should I learn i mean the basics ,starter framework or programming language! please
@jarad4621
3 ай бұрын
Same question I got Gemini to write a whole learning and study plan to get me to readiness for this kind of thing I suggest trying that with gpto maybe?
@morrazzo4432
3 ай бұрын
@@jarad4621 oh shit!! I'm sitting duck! I forgotten myself asking this to chatgpt 😭 Thanks for suggestions ! Have a good day homie!
@mrGapMan1
3 ай бұрын
Hey! the video is Ai. Isn't it?
@augustus6660
3 ай бұрын
Omg I think it is hahahah. The jacket's zipper slides around horizontally
@jarad4621
3 ай бұрын
@@augustus6660not seeing it if it is is very well done prob Heygen
@jarad4621
3 ай бұрын
Now that I think about it it's always the same position but even if it's Heygen avatar doesn't matter to me, we still get the value ?
@augustus6660
3 ай бұрын
Oh, dont get me wrong. This is the best content on this topic ive ever found. And tbh i think its only in the intro. Too much movement elsewhere
Пікірлер: 43