![Priya Dwivedi](/img/default-banner.jpg)
- Видео 10
- Просмотров 5 734
Priya Dwivedi
Канада
Добавлен 16 дек 2011
Hi, I'm Priya Dwivedi, founder of Deep Learning Analytics and ML enthusiast! I specialize in building custom machine learning applications through my consultancy.
Lately, I've been fascinated with generative AI and its possibilities. I started this RUclips channel to showcase my hands-on experiments and help other developers learn.
On this channel, you'll see me tinkering with the latest AI models and building applications. I'll walk through my approaches, technical implementations, and lessons learned through easily digestible commentary and visuals.
My goal is to make applied AI more accessible. My hope is that viewers not only enjoy the projects, but also gain skills to start building their own.
So subscribe if you want to learn by
doing and build your skills in applied machine learning and generative AI!
Lately, I've been fascinated with generative AI and its possibilities. I started this RUclips channel to showcase my hands-on experiments and help other developers learn.
On this channel, you'll see me tinkering with the latest AI models and building applications. I'll walk through my approaches, technical implementations, and lessons learned through easily digestible commentary and visuals.
My goal is to make applied AI more accessible. My hope is that viewers not only enjoy the projects, but also gain skills to start building their own.
So subscribe if you want to learn by
doing and build your skills in applied machine learning and generative AI!
Chat with API using a Groq LLM Agent
Looking to level up your API skills? We're diving into Groq, the lightning-fast LLM API, to create an intelligent agent that can tap into the OMDb database. This project is your gateway to building agents that can interact with real-world data sources. By the end of this video, you'll have the foundation to create all sorts of API-powered bots!
Udemy Course to learn about Agents:
www.udemy.com/course/building-intelligent-agents-with-llms/?couponCode=F6FE0CCFC20A9CFC7490
Streamlit Movie Agent: priya-dwivedi-movie-agent.streamlit.app
Github Code: github.com/priya-dwivedi/groq_movie_agent
Udemy Course to learn about Agents:
www.udemy.com/course/building-intelligent-agents-with-llms/?couponCode=F6FE0CCFC20A9CFC7490
Streamlit Movie Agent: priya-dwivedi-movie-agent.streamlit.app
Github Code: github.com/priya-dwivedi/groq_movie_agent
Просмотров: 109
Видео
Intelligent LLM Agents - New Udemy Course at 50% off
Просмотров 1652 месяца назад
I have released a new course on Udemy focused on building LLM agents. Use this course to : Master LLM Agents for Advanced Reasoning, Coding Assistants, Research Assistants, and More through Hands-On Code I am sharing here a couple of videos from this course that covers the basics of LLM Agents Get 50% off on the course. Use this link: www.udemy.com/course/building-intelligent-agents-with-llms/?...
Getting Gemma-7B working on Colab
Просмотров 5096 месяцев назад
This video covers how you can get the Gemma 7B model working on Colab Medium Blog on this project: generativeai.pub/the-dawn-of-gemma-googles-gift-to-the-ai-community-c9bba833a9b2 Colab Code: colab.research.google.com/drive/1YBNfnEUkeo6EVeLeu4btAFCGZrXxSETZ?usp=sharing Contact me: docs.google.com/forms/d/1exV3sAuJiemsqzAthGgv9PJ6c1hvZchLyD8jb2kekyY/edit Hit subscribe to get more videos like thi...
The AI Symphony: Building a Multi-Agent Pipeline with CrewAI
Просмотров 3396 месяцев назад
This video covers how you can use the CrewAI framework to build a multi agent pipeline Medium Blog on this project:generativeai.pub/the-ai-symphony-testing-crewais-multi-agent-system-for-complex-apps-b377d4f5e6b8 Github Code: github.com/priya-dwivedi/generative-ai/tree/main/assignment_helper Contact me: docs.google.com/forms/d/1exV3sAuJiemsqzAthGgv9PJ6c1hvZchLyD8jb2kekyY/edit Hit subscribe to g...
Making Chatbot Interactions Deeply Personal with Smart Memory
Просмотров 319 месяцев назад
In this video, I demo smart memory! A personalized memory that is created over a course of a conversation and can be used for the future.
How to extract tables and equations from books and papers correctly and chat with them? With code!
Просмотров 39511 месяцев назад
This video shows how to use the Nougat Model from Meta AI to parse research papers correctly. The output from the paper is returned in Latex format. I then use Langchain to detect and summarize all the tables in the source paper. Link to Nougat Paper: arxiv.org/pdf/2308.13418.pdf Link to Colab code: colab.research.google.com/drive/1oC7jK8UMEYRDAEPevn5VQweadjN4WyeW?usp=sharing Link to Medium Blo...
How to chat with an LLM using images? Share images of food, travel.. Get recipes and trip planning,
Просмотров 31711 месяцев назад
Open Source LLMs are soon going to be multi-modal. GPT4 technically has that capability but that version is not yet released. Idefics is an open-source LLM that you can upload images to and chat with them. This model has the amazing capability to identify food items from images and give recipes. Share photos of travel destinations and get advice on trip planning. Check the video to see my exper...
Can AI really build entire software with a 1 line prompt! See MetaGPT in action. Real Examples!
Просмотров 1,9 тыс.Год назад
Introducing MetaGPT - a framework that allows multiple chat agents to work together to build entire repos with good documentation. I have discussed here how to install MetaGPT and shared examples. Link to Paper: arxiv.org/pdf/2308.00352.pdf Link to Github repo: github.com/geekan/MetaGPT Follow me on LinkedIn: www.linkedin.com/in/priyanka-dwivedi-6864362 Hit subscribe to get more videos like thi...
Generate stunning images for free on colab using Stable Diffusion XL
Просмотров 159Год назад
You can now run Stable Diffusion XL on free Colab to get stunning images. Stable Diffusion XL is a serious enhancement over the previous version of previous versions of stable diffusion. Stable Diffusion XL now comes with a Refiner which is an optional second stage for further reducing noise and enhancing the output quality. The Colab code includes instructions to run this with and without the ...
Chat with Github repo using Langchain and DeepLake
Просмотров 1,8 тыс.Год назад
Hey Everyone, This video builds a Github Repo chatbot. The bot can be pointed to any repo and you can chat with the repo getting useful code snippets in seconds. The Streamlit app is hosted on the link below. chat-with-code-pdwivedi.streamlit.app/ I have used DeepLake as the Vector Store, Langchain to do the Conversation Retrieval, and Streamlit for the front end. Try it and share any feedback ...
Excellent video. Gonna experiment with this for other document types..
Can you please tell what to change in deeplake path as it is showing InvalidTokenException: Token is invalid. Make sure the full token string is included and try again.
I'm getting this error InvalidTokenException: Token is invalid. Make sure the full token string is included and try again. please help to solve this
Same here, is your problem solved now? If yes then how?
@@beenadwivedi2307 - I fixed this. The deeplake token I used had expired. Can you retry please and let me know if the issue persists
i am getting the Same Error : if you solve this please let me know
@@darshjain3133 - One of my requirements was out of date. Fixed that now. Can you please retry?
@@priya-dwivedi It's Still not Working :/ Maybe You Try using Chromadb for Storing the Data instead of Deeplake
but how to make a bot which will review all sort of codes after taking this as input?
Thanks for walking us through a useful multi-agent scenario in a way that is easy to follow. More CrewAI and AutoGen content please.
This is awesome Priya, exactly what I want to build next. Ive been interested in CrewAI but I didnt realize it had this pipelining capability, some of the other multi-agent frameworks can only handle the control flow for moderately complex tasks and need for the dev to define the overarching control flow across tasks, this looks like just what Ive been looking for, but there are so many tools out there not enough time in the day to try them all so I am grateful to you for putting this together, will like/share/subscribe!
Let's go!! this is amazing~ thank you!
This stopped running and all the codes and blogs related to understanding codebases with langchain have been removed, what's going on?
Hey, Thanks you for pointing that out. I checked that the DeepLake library had some changes and I updated the code to resolve the error. The link to Streamlit demo is: chat-with-code-pdwivedi.streamlit.app/ The Github code repo is the same Hope it works for you now.
Legit information thanks for sharing I’m going to play around with your repo I was literally thinking of building this exact application and it looks like you’ve built it better than I could. I’ve subscribed looking forward to more of your content.
Great content Priya. Thanks for your efforts!
🤣 P R O M O S M
Good Priya...please do go on creating multi agents...
Great video! A quick question - Noticed that the number of tokens was about 30k after removing the refrences. However the size of the tokens used in the langchain call was about 10k. How did it go from 30k to 10k? Thank you!
Good question. The len of content without references is 30k. length checks the total length of the string. The number of tokens for gpt3.5 is ~10k. A token is a subword which is typically 3-4 characters long.
Thank you
Are you looking for work ?
@@user-mx2vv7il2s idont get this working for my own document, some exception it lands up into File "/usr/local/lib/python3.10/dist-packages/transformers/generation/utils.py", line 2394, in _sample model_inputs = self.prepare_inputs_for_generation(input_ids, **model_kwargs) TypeError: BARTDecoder.prepare_inputs_for_inference() got an unexpected keyword argument 'cache_position'
Keep up the good videos:) Subscribed
🎯 Key Takeaways for quick navigation: 00:14 🤖 MetaGPT is a recent development that uses chatbots to create a multi-agent environment for collaborative tasks, including building software and forecasting models. 02:04 🧠 Clear personas, standardized outputs, and a shared messaging platform are used in MetaGPT to prevent agents from producing incorrect or erratic results. 03:00 ⚙️ Setting up MetaGPT involves installing a JavaScript package and configuring a YAML file with OpenAI and server API keys. 05:03 🎮 MetaGPT can be instructed to build complex applications, such as a tic-tac-toe game controlled via a command-line interface. 06:27 📝 MetaGPT not only generates code but also provides detailed documentation, making it useful for both designing and implementing software. 08:03 ⏲️ MetaGPT can be used for machine learning tasks, such as building time series forecasting models, with generated code as a starting point. 10:01 📈 While MetaGPT provides a good starting point, users may need to make small adjustments to the generated code for specific requirements. Made with HARPA AI
the same/similar examples given by AI RUclips content creators - you could have done differently by at least taking simple student-level projects.
This is helpful, Thanks Priya!
Thanks for sharing!
That's a wonderful demo @priya-dwivedi👏. It is interesting and exciting to see how Gen AI is evolving so rapidly. Apparently, the research into Gen AI is also seeing a large growth nowadays, even expanding on the multi-modalities front. It makes me wonder how soon we are going to see a model or architecture powerful and accessible as stable-diffusion that can generate films or videos fluidly based on a script or a series of prompts 😄.
Hey Krishna! Nice to talk again. I agree with you that the scope of the field is expanding very fast. I think designers are already using stable diffusion to craft smaller continuous scenes like for video games. But I won't be surprised if in a few years we have an AI generated movie.