🔴 This Agentic AI Workflow Will Take Over 🤯 Algorithm + Papers Explained
ฝัง
- เผยแพร่เมื่อ 19 มิ.ย. 2024
- #ai #llm #aiagents #agentic
What Language Model To Choose For Your Project? 🤔 LLM Evaluation:
• What Language Model To... : evaluation of Hugging Face models
Please subscribe to support this channel :)
Explanation of the papers and algorithms of LLM agents in the Agentic AI systems (see timestamps below) using the following concepts and papers:
Iterative feedback and refinement for LLM agents + evaluation:
Madaan et al. 2023. SELF-REFINE: Iterative Refinement with Self-Feedback.
The Reflexion algorithm explained in the paper:
Shinn et al. 2023. Reflexion: Language Agents with Verbal Reinforcement Learning
Automatic API calls system based on LlaMA + evaluation from the paper:
Gorilla: Large Language Model Connected with Massive APIs
HuggingGPT agentic system using Hugging Face, from the paper:
Shen et al. 2023. HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Face
Chain of Thought + evaluation on GSM8K tests from the paper:
Wei et al. 2022. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Agentic LLM + human customizable system for making applications and from the paper:
Wu et al. 2023. AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation
ChatDev agentic system for software development + its evaluation from the paper:
Qian et al. 2023. Communicative Agents for Software Development
Video key concepts:
00:00 Intro Agentic AI
00:40 Zero-shot use of LLMs
01:06 self-refine, iterative feedback and agentic refinement algorithm
02:16 Self-refine evaluation with ChatGPT and GPT4
02:29 Reflection method
02:33 Reflexion with verbal reinforcement learning and its evaluation
03:18 Explanation of the Reflexion algorithm and system design
04:26 Gorilla agentic system using tools and generating API calls
04:46 Gorilla evaluation with GPT-4 and Claude + evaluation on document retriever
05:09 Planning in agentic systems + Andrew NG takeaway message
05:23 Chain of Thought prompting for LLM step-by-step reasoning problems
05:36 Evaluation of chain of thought vs standard prompting of PaLM, Codex, GPT, LaMDA, UL2
05:Evaluation of chain of thought prompting on GSM8K tests + examples of chain of thought
06:03 HuggingGPT as LLM controller agentic system
06:23 HuggingGPT example of multi-modal tasks
06:34 HuggingGPT task planning, model selection, task execution, response generation
07:12 Multiagent collaboration: CrewAI and AutoGen
07:29 AutoGen agentic system for applications with customizable and conversable agents
08:05 AutoGen applications and use cases (math, ALF, multi-agent coding, group chat)
08:27 ChatDev agentic system for software development
Related concepts and key terms:
#ai #agentic #agentic ai #llm #chatdev #autogen #crewai #hugginggpt #huggingface #self-refine #gpt4 #reflexion #gorilla #api #planning #gsm8k #multimodal #multiagent #ALF
--Don’t forget to subscribe and watch these related videos:
Transformer Language Models Simplified in JUST 3 MINUTES!
• Transformer Language M...
This Is How Exactly Language Models Work in AI - NO background needed! • This is how EXACTLY La...
The Concept of Backpropagation Simplified in JUST 2 MINUTES! --Neural Networks • The Concept of Backpro...
/ @analyticscamp - วิทยาศาสตร์และเทคโนโลยี
Some of you asked for AI agents in action; here's a video with code to use 100% free local AI agents: th-cam.com/video/XkS4ifkLwwQ/w-d-xo.html
Very nice video! I liked the explanation of the Reflexion algorithm. Please do more of ai agents.
Thanks, will do!
nice video !
Thank you! Cheers!
Great content !! thx ! and yes agentic is my exploration for months... And it's still very difficult to masterize the agentic flow.. sometimes it's going out of limits , i didn't figure out yet how to fix that (with both autogen and crewai)... maybe a topic for a future video ?
Great suggestion. Yes it can be challenging at times :)
Update: Here's the video you requested: th-cam.com/video/XkS4ifkLwwQ/w-d-xo.html
Nice video
Thanks for watching :)
The video is very informative but would be great to see the agent power in action
Thanks for watching :) Here's AI agents in action: th-cam.com/video/XkS4ifkLwwQ/w-d-xo.html
more agents are all you need bro, dont treat AI agent as human, think of it as the cell in your body (maybe this is bad analogy), so we need hundreds of them at least to perform certain function
Thanks for your comment! Maybe you are right!
Yeah, Pat and Mat, thats from Czech studios :-)
Thanks for watching :)
Interesting video, good work, but it's very silent, I had to more than double times increase volume to get output like from other random video, somethings wrongs with the sound or style how you recorded ;)
Thanks for your feedback. Sorry about the sound quality; I'll try to fix it in the next videos. Stay tuned :)
Good video! Could you make a comparative with Mamba structure?
Great suggestion! I'll see :)
I like your style, content and voice. Just earned a sub.
Welcome aboard :)
Same! 🔥🔥🔥
Nice video, nice channel, nice voice, nice pronuntiation. Clear to understand ;)
Glad you think so and thanks for watching :)
unwatchable with headphones.
Apologies for the sound quality. Please watch this video using a speaker. Thanks :)
you know this is made by human, and this human is so busy with work... so this video have mono audio at left speaker only.
don't bother busy people doing their busy thing. you are asking too much.
Thanks for your feedback! I used to record with Zoom but I switched to this new mic; will need to adjust it as you suggested :)
@@analyticsCamp can't wait another explanatory vids like this.
Thanks for your feedback!
Oh come on. What if I tell you, you have been living your life wrong, and there is a much better way...
LOL 😂 Thanks for watching :)
gorilla seems like the holy grail. it's strange that it seems to have reached a standstill in development.
Thanks for watching. Their newest development is GoEX, their paper (below reference) talks about how humans can supervise autonomous LLMs. Here's the paper ref:
Patil et al., 2024. GoEX: Perspectives and Designs Towards a Runtime for Autonomous LLM Applications.
@@analyticsCamp Groovy, thanks for that!
👍