![Automata Learning Lab](/img/default-banner.jpg)
- 161
- 283 322
Automata Learning Lab
Registrace 1. 02. 2021
What up folks! My name is Lucas, I am an AI Engineer and in this channel I want to explore topics like AI, automations and how to build some cool tools!
A Quick Introduction to LangGraph for Building Agentic Workflows Using GPT-4o-Mini
In this video, we'll talk about the basics of LangGraph, a framework for building powerful agentic workflows based on large language models. We'll use OpenAI's gpt-4o-mini as the base model for our examples and walk through the basics of langgraph and some interesting concepts and ideas around agentic systems!
Disclaimer: (Yes my face did get cut off in the video...I only realized that after having recorded...living and learning folks...)
📚 Chapters:
00:00 - Introduction to LangGraph: Welcome and overview of the video
00:05 - What is LangGraph?
00:10 - Overview of Agentic Systems
1:13 - The Agent Loop: Explanation of the agent loop concept
5:53 - Key Components of Agentic Systems: Essential elements like memory, planning, and tool calling
6:42 - Modeling Agentic Workflows as graphs
7:53 - Core Principles of LangGraph: Controllability, human-in-the-loop, and streaming-first design
11:22 - Components of LangGraph: Explanation of nodes, edges, and states in LangGraph
13:09 - Example Implementation: Step-by-step construction of a basic agent workflow in LangGraph
19:30 - Nodes in LangGraph: Functionality, execution, and special nodes
21:22 - Edges in LangGraph: Different types of edges and their purposes
23:09 - Integrating Edges: Practical example of adding edges to the graph
27:46 - Inspecting Final State: Reviewing the final output and state of the graph
29:33 - Agent Loop Outline: Summary of the basic agent loop
31:44 - Full Example - Research Agent: Detailed demo of a research agent use case
42:38 - Agentic Patterns: Overview of agentic patterns like plan and execute, supervisor, and multi-agent workflow
45:13 - Conclusion: Final thoughts and encouragement to explore LangGraph and related resources
🔗 Links:
- Subscribe!: czcams.com/channels/u8WF59Scx9f3H1N_FgZUwQ.html
- Tiktok: www.tiktok.com/@enkrateialucca?lang=en
- Twitter: LucasEnkrateia
- LinkedIn: www.linkedin.com/in/lucas-soares-969044167/
- AI project tracker from Ben's Bites: bensbites.beehiiv.com/subscribe?ref=CoXhc4I0c0
Support the Channel!
- Buy me a cup of coffee: tr.ee/7tYsD-tUu2
- Paypal: paypal.me/lucasenkrateia?country.x=PT&locale.x=pt_PT
Disclaimer: (Yes my face did get cut off in the video...I only realized that after having recorded...living and learning folks...)
📚 Chapters:
00:00 - Introduction to LangGraph: Welcome and overview of the video
00:05 - What is LangGraph?
00:10 - Overview of Agentic Systems
1:13 - The Agent Loop: Explanation of the agent loop concept
5:53 - Key Components of Agentic Systems: Essential elements like memory, planning, and tool calling
6:42 - Modeling Agentic Workflows as graphs
7:53 - Core Principles of LangGraph: Controllability, human-in-the-loop, and streaming-first design
11:22 - Components of LangGraph: Explanation of nodes, edges, and states in LangGraph
13:09 - Example Implementation: Step-by-step construction of a basic agent workflow in LangGraph
19:30 - Nodes in LangGraph: Functionality, execution, and special nodes
21:22 - Edges in LangGraph: Different types of edges and their purposes
23:09 - Integrating Edges: Practical example of adding edges to the graph
27:46 - Inspecting Final State: Reviewing the final output and state of the graph
29:33 - Agent Loop Outline: Summary of the basic agent loop
31:44 - Full Example - Research Agent: Detailed demo of a research agent use case
42:38 - Agentic Patterns: Overview of agentic patterns like plan and execute, supervisor, and multi-agent workflow
45:13 - Conclusion: Final thoughts and encouragement to explore LangGraph and related resources
🔗 Links:
- Subscribe!: czcams.com/channels/u8WF59Scx9f3H1N_FgZUwQ.html
- Tiktok: www.tiktok.com/@enkrateialucca?lang=en
- Twitter: LucasEnkrateia
- LinkedIn: www.linkedin.com/in/lucas-soares-969044167/
- AI project tracker from Ben's Bites: bensbites.beehiiv.com/subscribe?ref=CoXhc4I0c0
Support the Channel!
- Buy me a cup of coffee: tr.ee/7tYsD-tUu2
- Paypal: paypal.me/lucasenkrateia?country.x=PT&locale.x=pt_PT
zhlédnutí: 734
Video
Building ChatGPT's App Ideas with Claude 3.5 Sonnet & Artifacts
zhlédnutí 1,2KPřed 19 hodinami
In this video we'll build 3 app ideas (I started with 10....) generated with ChatGPT using Claude 3.5 Sonnet and Artifacts. 📚 Chapters: 00:00 - Introduction: Trying New App Ideas with AI 00:08 - Overview: Generating and Implementing App Ideas 00:34 - Strategy: Using CHPT for Idea Generation 01:06 - Implementation Plan: From Idea to Prototype 03:01 - Detailed Workflow: Iteration and Validation 0...
Quick Introduction to Fast Tool Calling with Groq and Llama3
zhlédnutí 284Před dnem
In this video we'll look at tool calling with Groq's fine tuned Llama 3 model specialized in for function calling. We'll conduct some simple experiments profiling Llama 3 models, comparing Llama3 70B and 8B models for latency given a simple prompt. At the end we'll also take a quick look at the super cool demo by Groq that combines fast audio transcription with Whisper Large v3 and Groq's optim...
How to Use Claude 3.5 Sonnet & Artifacts for Studying, Learning & Research
zhlédnutí 4,7KPřed 21 dnem
#claude #llm #ai Some cool use cases for Claude 3.5 Sonnet & Artifacts for studying, learning and doing research. 📚 Chapters: 00:00 - Introduction to the Video and Goals 00:38 - Summarizing a Paper with Claude AI 01:21 - Turning a Paper into an Interactive Dashboard App 03:01 - Verifying the Data in the Interactive Dashboard 04:03 - Iterating on Interactive Dashboard with Feed Forward Network E...
Building Interactive Apps with Claude 3.5 Sonnet & Artifacts
zhlédnutí 11KPřed měsícem
#claude #artifacts #llm #ai Let's take a look at awesome use cases for Claude 3.5 Sonnet including: creating interactive interactive quiz apps directly from PDFs, creating interactive mind maps, creating dynamic flowcharts that can dynamically react to selection of parts of the code, and more! 📚 Chapters 00:00 - Exploring Interactivity And Prompting in Claude 3.5 Sonnet 00:35 - Performance of C...
From Text To Visual Diagrams with LangChain, Pydantic & ChatGPT
zhlédnutí 451Před měsícem
In this we'll learn how to use models like ChatGPT to transform text into clear, visual Knowledge Graphs, leveraging LangChain & Pydantic. We'll walk you through an easy-to-follow Jupyter Notebook, detailing the imports, classes, and functions needed to automate the process using libraries like Pydantic and OpenAI. We even create some cool visualizations, showing step-by-step how to turn a simp...
Automating High Quality Quizzes from Video with Whisper & GPT-4o
zhlédnutí 251Před měsícem
Learn how to effortlessly create a high-quality quiz from a video in this step-by-step tutorial! We'll walk you through transforming video content into engaging multiple-choice quizzes by converting video to audio, generating transcriptions, and using AI to draft questions. We'll cover essential tools like MoviePy, OpenAI Whisper, and ChatGPT-4, as well as provide tips for crafting well-structu...
Automating The Setup of a Github Repo with ChatGPT, LangChain & Pydantic
zhlédnutí 510Před 2 měsíci
In this video, we dive into automating the process of setting up a GitHub repository by using Pydantic and LangChain to create an extraction chain. We'll guide you through the step-by-step setup, starting from importing necessary libraries to configuring a Pydantic object, and converting it into a function that integrates with the GPT-4 model. Like & Subscribe! :) Cheers. 📚 Chapters: 00:00 - In...
Understanding Scientific Papers With Prompt Engineering & The ChatGPT API
zhlédnutí 462Před 2 měsíci
I this video we'll look at a fun and insightful approach using advanced models like ChatGPT to dive deep into understanding scientific papers. Join us as we explore how to implement a prompt engineering pipeline based on Tim Roughgarden's lecture notes from Stanford. We'll dive into the key questions that help evaluate a paper, such as the problem it addresses, its primary contributions, and mo...
Prompt Engineering for Code Generation
zhlédnutí 573Před 2 měsíci
In this video, we delve into a fascinating code generation use case for prompt engineering. We tackle the task of creating a simple Python function that performs operations between two numbers using large language models. By setting up clear test cases, we evaluate the effectiveness of the generated code and ensure its reliability. We also touch on the iterative process of refining and expandin...
Structured Outputs with Pydantic & OpenAI Function Calling
zhlédnutí 581Před 2 měsíci
In this video, we delve into the fascinating realm of open AI function calling and getting structured outputs from Large Language Models! We start by explaining what function calling is and how OpenAI implements this feature. We demonstrate how to use Pydantic to create structured data types combined with OpenAI's function calling, leading to the generation of a fully functional quiz from an ar...
A Simple Prompt Engineering Experiment Template for Text Summarization
zhlédnutí 507Před 2 měsíci
In this second video of our Prompt Engineering series, lets walk you through a systematic framework for building and experimenting with prompts to achieve better results. We'll go through a text summarization example including task definition, setting up an evaluation metric using GPT-4 as a judge to score the quality of the summaries using a 4 point criteria, create a few prompt candidates the...
A Quick & Practical Introduction to Prompt Engineering
zhlédnutí 511Před 3 měsíci
A Quick & Practical Introduction to Prompt Engineering
Learning with AI | Building a Quiz App with GPT-4 and LangChain
zhlédnutí 792Před 4 měsíci
Learning with AI | Building a Quiz App with GPT-4 and LangChain
Getting Started with MultiOn the Best AI-Powered Web Browser Agent
zhlédnutí 1,6KPřed 4 měsíci
Getting Started with MultiOn the Best AI-Powered Web Browser Agent
Fast PDF Summarization & Semantic Zoom Using Groq & Mixtral 8x7b
zhlédnutí 801Před 4 měsíci
Fast PDF Summarization & Semantic Zoom Using Groq & Mixtral 8x7b
Automating Creation of Anki Flashcards from ChatGPT Conversations using GPT-4
zhlédnutí 694Před 5 měsíci
Automating Creation of Anki Flashcards from ChatGPT Conversations using GPT-4
My Failed Experiments with GPT-4 Vision Localization Capabilities for Anki Flashcards
zhlédnutí 158Před 5 měsíci
My Failed Experiments with GPT-4 Vision Localization Capabilities for Anki Flashcards
Using GPT4 Vision & LangChain to Query Memories from a Video
zhlédnutí 535Před 5 měsíci
Using GPT4 Vision & LangChain to Query Memories from a Video
Building a Simple Quick Capture Agent with AutoGen
zhlédnutí 594Před 5 měsíci
Building a Simple Quick Capture Agent with AutoGen
An Introduction to LLM Agents | From OpenAI Function Calling to LangChain Agents
zhlédnutí 13KPřed 6 měsíci
An Introduction to LLM Agents | From OpenAI Function Calling to LangChain Agents
LLM Agents that Can Learn Facts and Skills | AutoGen
zhlédnutí 985Před 6 měsíci
LLM Agents that Can Learn Facts and Skills | AutoGen
5 rules for Building LLM Agents with AutoGen
zhlédnutí 851Před 6 měsíci
5 rules for Building LLM Agents with AutoGen
Building LLM Agents That Generate Tailored Workout Routines with LangChain
zhlédnutí 1,3KPřed 6 měsíci
Building LLM Agents That Generate Tailored Workout Routines with LangChain
Building LLM Agents in 3 Levels of Complexity: From Scratch, OpenAI Functions & LangChain
zhlédnutí 9KPřed 7 měsíci
Building LLM Agents in 3 Levels of Complexity: From Scratch, OpenAI Functions & LangChain
Getting Started with OpenAI GPT Builder
zhlédnutí 1,4KPřed 8 měsíci
Getting Started with OpenAI GPT Builder
Building a PDF Translation App with ChatGPT
zhlédnutí 1,1KPřed 8 měsíci
Building a PDF Translation App with ChatGPT
Quick Automations with Python | Creating PDFs from Your Clipboard
zhlédnutí 192Před 8 měsíci
Quick Automations with Python | Creating PDFs from Your Clipboard
3 Ways To Set Up Llama2 Locally | Llama Cpp, Ollama, Hugging Face
zhlédnutí 3,7KPřed 9 měsíci
3 Ways To Set Up Llama2 Locally | Llama Cpp, Ollama, Hugging Face
Building an LLM Agent for Github with LangChain
zhlédnutí 1,8KPřed 9 měsíci
Building an LLM Agent for Github with LangChain
2:56
Excellent presentation, very simple & clear articulation for a highly complex topic
@@bigdlamz thanks!!! 😊
Thank you! I love the videos from this channel. They're always so informative and helpful! Could you explain how LangGraph, AutoGen, Semantic Kernel, LlamaIndex and Haystack differ in their approach to creating agentic workflows? Specifically, how do these tools integrate with each other, and what are the unique advantages of using each tool individually or together in a project? Additionally, could you discuss the roles of LangSmith and LangTrace AI in this context and how they compare or complement the other tools?
Havent' used LangTrace. LangSmith is an evaluation and monitoring platform created on top of langchain so you can inspect everything that happens between chains which are these building blocks of processing that happens within these agentic workflows. LangGraph allows you to create workflows that mix determinic processing of inputs and outputs along with LLM processing (less deterministic). AutoGen takes a multi-agent approach that leverages some communication principles and concepts like 'conversation programming' to allow for implementing complex agentic workflows to solve complex tasks, is a different approach than this gtraph based approach that langgraph takes. Llama index was previously a rag focused platform but now they are also steering towards agentic workflows as the field sees its the best approach to actually tackle complex problems. Platforms like LangSmith are going to be key because anyone working with these systems sees that the issue always goes back to evaluation practices, monitoring etc....that's it! Cheeers! :) (thanks for watching!)
hey. nice video idea! i don't understand how you then can create a working app ? do you have to take the created code and put it into android studio?
If you want something already like structured as an actual app, try out gpt-engineer. IT will spit out the code already in a github repo. However, if you're working with Claude 3.5 sonnet and artifacts you can download the code generated, then ask it to create the folder structure with steps to execute the created app. It will spit out all the steps needed plus the packages to install etc....from there its like super easy to set upyour working app locally! Hope that helps! :) cheers!
@@automatalearninglab thank's a lot!:). i'm a total beginner when it comes to these kinda things. can't even code. but maybe one day you'll make a beginners guide called "from sonnet -to- finished local app":) cheers, Mat (from Germany)
@@bluetensee haha that's a cool idea!
Eu pessoalmente gosto de ambos o Claude e o ChatGPT, mas nada bate um Lucas Soares 😍😍😍
hahahah thank you! ;)
If you would have budget to only have one paid AI. Would you invest in Claude our ChatGPT?
I think as a developer I would choose Claude, as a non developer I would go with ChatGPT. :)
Do you have pro- subscription? I tried the exact prompt but it didn't render any app for me. Instead, gave me instructions on how to do it...
i do have it, and I have Artifacts feature enabled. Check it under settings if you also have that enabled.
Nice video. Like that you walk through in real time... including showing how much co-pilot currently sucks.
@@IdPreferNot1 hahaha yeah that was not planned but I’m sure they’ll update it
What size of Llama3 are you using? At which point it starts not being useful? I really struggle to decide which size to use. Great video!
@@AlexandreAugustin for small scale tasks use the 8B, and for efficient tool calling 70b. Is really good. I am using both in this video
Great video with deep insights and high levels. It will be good to combine a second video with runnable codes to visually see the results.
Oh. nice! Great idea! :)
Love this man... our projectlife foundation is rehumanising mother and child care through tech for the doctors, would love to discuss what we are doing with you and find a possible collaboration, i am sure you will love what we are working on. How can we connect?
That's awesome! Hit me up on LinkedIn! :) www.linkedin.com/in/lucas-soares-969044167/
Sure thanks@@automatalearninglab
🎯 Key points for quick navigation: 00:00 *🧩 CLA 3.5 and Artifacts Overview* - CLA 3.5 is a new model by Anthropic with superior performance metrics compared to GP4. - Artifacts enable standalone interactive content alongside CLA conversations, enhancing dynamic coding experiences. - Example: Building interactive quiz apps from PDF content using CLA 3.5 and Artifacts. 01:51 *📝 Interactive Quiz App Development* - Using CLA 3.5 and Artifacts to create quizzes based on PDF content. - Features include generating multi-choice and open-ended questions dynamically. - Example: Iterative improvement of quiz app interface and functionality. 07:47 *🗺️ Interactive Mind Maps* - Generating draggable mind maps to visualize large language model use cases. - Enhancing readability and usability of generated mind maps through iterative adjustments. - Example: Exploring different visualization layouts and customization options. 11:11 *🔄 Flowchart Generation from Code* - Using CLA 3.5 and Artifacts to create flowcharts from Python code snippets. - Iterative development to refine flowchart visualization and interactivity. - Example: Building a dynamic interface linking code segments to flowchart nodes. 19:43 *📊 Interactive Data Visualizations* - Exploring CLA 3.5 capabilities in generating interactive data visualizations. - Utilizing artifacts to create visual representations directly from structured data files. - Example: Demonstrating real-time customization and usability of interactive data displays. 23:34 *📊 Data Simulation and Code Generation* - Demonstrates generating simulated data and code within Cloud Sonnet, - Highlights creating a table with simulated plant growth and soil pH levels, - Initiates data organization into CSV format and visual dashboard creation. 25:07 *📈 Building Interactive Dashboard* - Shows how to build an interactive dashboard to visualize data dynamically, - Integrates simulated data into a dashboard with adjustable parameters, - Illustrates plotting growth rate against soil pH for different vegetables. 26:47 *📊 Uploading and Visualizing CSV Data* - Guides uploading CSV data and creating an interactive dashboard, - Demonstrates visualizing CZcams statistics from a downloaded CSV, - Discusses issues with visualization layout and potential improvements.
@@v-for-victory nice
which SLM you would recommend for handling summaries in both English and French .thank you
Not aware of good ones for French but I would fine tune llama3.1 70b or 8B probably
its free?
hey, I'd like to sponsor you for my ai software - let me know how I can contact you to chat further
Love it, try to ask it for suggestions of design and usability. The results are so much beater than what you are geting. Hope it helps
@@carlosperezcpe Nice! Thanks! :)
Amazing
@@marcanttins tx!
Amazing
Amazing
Amazing
Amazing
@@marcanttins tx !
Amazing
@@marcanttins tx! :)
Amazing
@@marcanttins tx!
Some features my not work on the preview section. You have to add the code to the real application and run it
@@khanhnguyenxuan4042 yeah sure. I think here I was just trying to explore possibilities around the topic of learning etc
Thanks for the introduction for the application of ChatGPT with Anki!
Thank youuuu
Great stuff
Definitely hit my niche as a Claude 3.5 + Anki learner who wants to optimize my flashcards. Very informative and gives me a lot of ideas. Even just the idea of Claude's coding errors being more of a speed bump than a dead end. I finally have a reason to try Artifacts, thanks!
@@cameosisj6007 oh I’m glad! Thanks for watching!
Thank you for sharing your knowledge; it was highly informative. 👍
You're welcome!
Nice video!
Thanks! :)
Agreed. It’s a bit better than 4o bc it doesn’t split output code. And it can immediately show the code instead of copy-pasting to something like codepen to get a visual
Yep, the whole experience with artifacts plus great code generation plus this atomic control over the code snippets make this a great tool! I have some other ideas lined up for what to build with it! :) Thanks for watching!
Amazing
@@carlosbaraldi124 thanks!
how do i create virtual environment in linux could you please tell how cause i have to do this project
easy, just download anaconda or miniconda, then run the cmd to create the environment like: conda create -n env-name python=<python version> then activate the env with 'conda activate env-name' and now you can install packages with pip install <package> Hope this helps! :)
at 2:38 how did the window on the right just appear for you? Are you using a specific version of Claude 3.5 Sonnet? I am using the free version of Claude.
THat's the artifacts feature that you have to enable! :)
This drove me crazy too!!! Basically, in Claude you click on your name at the bottom left of the main window. Then click on "Feature Preview". When the window pops up, you will see the Artifacts artical/preview appear and in the last third of that window, you will see an "Artifacts" slider switch that you can turn on and off. That's it! I found it by sitting through a bunch of videos and in one video it was quicky mentioned so I got lucky, but really it could be mentioned a lot more, especially when these video makers are referencing the Artifacts feature!
@@Mr_Kweeba agreed! 😎
What is your plan on claude ai ? I see you have unlimited messages to use
I use the pro version!
Can it give summarization for any research documents?
Love how other people use LLM's creatively. The code/flowchart example was amazing! Thanks Lucas!
Thank you @duhai1836!
@automatalearninglab I am facing an error in running this script help me out. I had done exactly what you told to do. File "C:\Users\User\Downloads\voice_analyzer_kyro\voice_analyzer.py", line 27, in <module> model = whisper.load_model("base") ^^^^^^^^^^^^^^^^^^^^^^^^^^ File "c:\Users\User\Downloads\voice_analyzer_kyro\.env\Lib\site-packages\whisper\__init__.py", line 154, in load_model model.set_alignment_heads(alignment_heads) File "c:\Users\User\Downloads\voice_analyzer_kyro\.env\Lib\site-packages\whisper\model.py", line 251, in set_alignment_heads mask = torch.from_numpy(array).reshape( ^^^^^^^^^^^^^^^^^^^^^^^ RuntimeError: Numpy is not available
Thank you,,,
Your welcome :)
AAAAH coisa mais linda que eu já vi na minha vida (o youtuber) 😍😍😍 (mas o modelo é fixe também)
🤣🤣🤣🤣rapariga tu não disfarça 🤣🤣🤣☺️
@@automatalearninglab sua belezura que não disfarça 😘😘😘
@@beatrizbelbut4862 hahaha
I loved the first half of the video talking about agents from scratch , this is very educational and not highly covered . all stuff use llama Index, Langchain and other frameworks which is my opinion is a hateful abstraction . with gpt-4o I think we will not even the cumbersome OpenAI function calling
I think function calling is definitely here to stay but I understand what you're saying about the hateful abstractions, I teach courses on these frameworks for OReilly Media and even for me is a pain to keep up with the complexiity and ever changing nature of these chaining abstractions.
Awesome video,
Thank you! :)
If you could choose an actionable o’reilly course to learn & begin earning money as a freelancer, what would you choose? Any course which teaches real life implementation.
I am inclined to suggest my courses for obvious reasons, but since the goal is to earn money as a freelancer, maybe my content would be a bit too much focused on like implementations rather than deployment. I don't know many courses from OReilly in that sense, however there are many youtubers with great content on stuff like that, like Jason AI and some other guys.
Can i get ur githhub repo link
Is it possible to also do speaker recognition? Do you have a video for it?
is the code available?
Amazing tutorial dude! Thanks 🙏
no worries!
this video is a goldmine of knowledge, thank you
kk thanks!
😍😍😍
hehe
Top !
Thanks!!
Great video
Thanks! :)