MIT Introduction to Deep Learning 6.S191: Lecture 4
Deep Generative Modeling
Lecturer: Ava Amini
*New 2024 Edition*

For all lectures, slides, and lab materials: http://introtodeeplearning.com​

Lecture Outline
0:00​ - Introduction
6:10- Why care about generative models?
8:16​ - Latent variable models
10:50​ - Autoencoders
17:02​ - Variational autoencoders
23:25 - Priors on the latent distribution
32:31​ - Reparameterization trick
34:36​ - Latent perturbation and disentanglement
37:40 - Debiasing with VAEs
39:37​ - Generative adversarial networks
42:09​ - Intuitions behind GANs
44:57 - Training GANs
48:28 - GANs: Recent advances
50:57 - CycleGAN of unpaired translation
55:03 - Diffusion Model sneak peak

Subscribe to stay up to date with new deep learning lectures at MIT, or follow us @MITDeepLearning on Twitter and Instagram to stay fully-connected!!

The Scrapy Beginners Course will teach you everything you need to learn to start scraping websites at scale using Python Scrapy.

The course covers:
- Creating your first Scrapy spider
- Crawling through websites & scraping data from each page
- Cleaning data with Items & Item Pipelines
- Saving data to CSV files, MySQL & Postgres databases
- Using fake user-agents & headers to avoid getting blocked
- Using proxies to scale up your web scraping without getting banned
- Deploying your scraper to the cloud & scheduling it to run periodically

✏️ Course created by Joe Kearney.

⭐️ Resources ⭐️
Course Resources
- Scrapy Docs: https://docs.scrapy.org/en/latest/
- Course Guide: https://thepythonscrapyplayboo....k.com/freecodecamp-b
- Course Github: https://github.com/orgs/python....-scrapy-playbook/rep
- The Python Scrapy Playbook: https://thepythonscrapyplaybook.com/

Cloud Environments
- Scrapyd: https://github.com/scrapy/scrapyd
- ScrapydWeb: https://github.com/my8100/scrapydweb
- ScrapeOps Monitor & Scheduler: https://scrapeops.io/monitoring-scheduling/
- Scrapy Cloud: https://www.zyte.com/scrapy-cloud/

Proxies
- Proxy Plan Comparison Tool: https://scrapeops.io/proxy-pro....viders/comparison/fr
- ScrapeOps Proxy Aggregator: https://scrapeops.io/proxy-api-aggregator/
- Smartproxy: https://smartproxy.com/deals/proxyservers/ips

⭐️ Contents ⭐️
⌨️ (0:00:00) Part 1 - Scrapy & Course Introduction
⌨️ (0:08:22) Part 2 - Setup Virtual Env & Scrapy
⌨️ (0:16:28) Part 3 - Creating a Scrapy Project
⌨️ (0:28:17) Part 4 - Build your First Scrapy Spider
⌨️ (0:55:09) Part 5 - Build Discovery & Extraction Spider
⌨️ (1:20:11) Part 6 - Cleaning Data with Item Pipelines
⌨️ (1:44:19) Part 7 - Saving Data to Files & Databases
⌨️ (2:04:33) Part 8 - Fake User-Agents & Browser Headers
⌨️ (2:40:12) Part 9 - Rotating Proxies & Proxy APIs
⌨️ (3:18:12) Part 10 - Run Spiders in Cloud with Scrapyd
⌨️ (4:03:46) Part 11 - Run Spiders in Cloud with ScrapeOps
⌨️ (4:20:04) Part 12 - Run Spiders in Cloud with Scrapy Cloud
⌨️ (4:30:36) Part 13 - Conclusion & Next Steps

🎉 Thanks to our Champion and Sponsor supporters:
👾 davthecoder
👾 jedi-or-sith
👾 南宮千影
👾 Agustín Kussrow
👾 Nattira Maneerat
👾 Heather Wcislo
👾 Serhiy Kalinets
👾 Justin Hual
👾 Otis Morgan

--

Learn to code for free and get a developer job: https://www.freecodecamp.org

Read hundreds of articles on programming: https://freecodecamp.org/news

A self-driving car in GTA 5.

For more information on this project, and how it all began from simple lane detection to deep learning, follow the full tutorial series here: https://pythonprogramming.net/....game-frames-open-cv-

Twitch.tv/sentdex
discord.gg/sentdex

Learn the essentials of working with AI in the cloud from @ExamProChannel. This comprehensive course covers the complete generative AI development lifecycle - from fundamental concepts through practical implementation, including prompt engineering, development tools, deployment, optimization, and advanced topics like RAGs and AI agents.

More course info: https://www.exampro.co/exp-genai-001

❤️ Try interactive AI courses we love, right in your browser: https://scrimba.com/freeCodeCamp-AI (Made possible by a grant from our friends at Scrimba)

⭐️ Contents ⭐️
- 00:00:00 Introduction
- 00:54:16 AI and ML Fundamentals
- 03:02:21 Gen AI Primer
- 03:32:55 Data and ML
- 03:47:56 LLM Basics
- 04:12:22 AI Powered Assistants
- 04:24:42 Env Setup
- 06:12:17 Prompt Engineering
- 07:00:25 WorkBenches and Playgrounds
- 07:44:09 Model as a Service
- 08:36:26 LLM DevTools and Workflow
- 11:52:07 AI Code Assistants
- 14:04:37 App Prototyping
- 17:21:06 Containers
- 18:12:43 Serving
- 18:19:51 AI Delivery Platform
- 19:40:45 GenAI Hardware
- 19:50:21 Framework
- 19:51:49 LLM Customization
- 19:52:35 SFT
- 19:56:25 Size Optimization
- 20:26:04 RAGS
- 22:21:19 Agents

🎉 Thanks to our Champion and Sponsor supporters:
👾 Drake Milly
👾 Ulises Moralez
👾 Goddard Tan
👾 David MG
👾 Matthew Springman
👾 Claudio
👾 Oscar R.
👾 jedi-or-sith
👾 Nattira Maneerat
👾 Justin Hual

--

Learn to code for free and get a developer job: https://www.freecodecamp.org

Read hundreds of articles on programming: https://freecodecamp.org/news

Introduction to Langchain Javascript Documentation. How to Create GPT-3 GPT-4 Chatbots that can contextually reference your data (txt, JSON, webpages, PDF) with embeddings . Brief Introduction into embeddings, vectorstorage options such as Pinecone, Chroma.

Consulting 1hr Call: https://cal.com/starmorphai/co....nsultingcall?duratio
Starmorph Website: https://Starmorph.com
NextJS Trained Bot: https://code.chat
Langchain JS Docs: https://js.langchain.com/docs/
Langchain Python Docs: https://langchain.readthedocs.io/en/latest/
OpenAI Embeddings Docs: https://platform.openai.com/do....cs/guides/embeddings
Zahid Open Source Langchain: https://github.com/zahidkhawaj....a/langchain-chat-nex
Instagram: https://instagram.com/starmorphdesign

Learn how to design and use databases. You will create a graphical Windows application with C# that can display and modify data from a mySQL database server.

You need Microsoft Windows to follow along.

✏️ @shadsluiter is your instructor for this course. He is a Professor Of Computer Science and Software Development at Grand Canyon University.

🔗 Professor Sluiter has more courses here: http://studycoding.org

⭐️ Contents ⭐️
⌨️ Course Outline
⌨️ Create a new database in mySQL
⌨️ How to write sql queries
⌨️ Create a GUI front end app
⌨️ Create a Data Access Object and connect to a database
⌨️ Query the database
⌨️ Add a picture box to display album art
⌨️ Insert new records into the database
⌨️ Create a second table and a foreign key
⌨️ Table joins part a
⌨️ Table joins part b
⌨️ Design with UML and ER diagrams
⌨️ Perform compound select actions
⌨️ Delete data from a table
⌨️ Add a video player to the app
⌨️ Extended Features

⌨️ (0:00:00) Introduction
⌨️ (0:04:33) Tools we’ll need
⌨️ (0:08:49) Creating a new database
⌨️ (0:12:16) Saving the data
⌨️ (0:16:04) How to fill in a table
⌨️ (0:20:26) Select star from albums
⌨️ (0:23:44) How to delete backticks
⌨️ (0:27:16) Intro to the series
⌨️ (0:34:40) How to delete the backticks
⌨️ (0:37:49) Building the database app
⌨️ (0:42:22) Start of the demo
⌨️ (0:46:33) Creating an album class
⌨️ (0:51:19) Adding two albums to the list
⌨️ (0:55:11) Connecting the database to the database
⌨️ (0:58:08) Where does the data source live?
⌨️ (1:04:42) Setting up the database
⌨️ (1:07:53) Running the app
⌨️ (1:10:37) How to create searches
⌨️ (1:13:55) Programming the new search button
⌨️ (1:18:12) The problem with parameter queries
⌨️ (1:21:56) Adding a picture to the screen
⌨️ (1:26:12) How to test the problem
⌨️ (1:30:45) Adding new records to the database
⌨️ (1:39:22) Adding an integer to the database
⌨️ (1:43:27) Adjusting the column titles
⌨️ (1:46:13) Execute non-query query
⌨️ (1:49:06) Creating a foreign key
⌨️ (1:51:47) Install mysql workbench
⌨️ (1:55:05) Connecting to the database
⌨️ (1:59:04) Configuring the tracks table
⌨️ (2:06:49) We have a problem with Mamp synchronization
⌨️ (2:10:01) Inserting data into the tracks table
⌨️ (2:13:08) Selecting items from the database
⌨️ (2:16:46) Changes to the definition of tracks
⌨️ (2:20:55) Fixing the problem with the binding source
⌨️ (2:24:12) Foreign keys and joins
⌨️ (2:27:36) Selecting from the table
⌨️ (2:31:15) Changing the query using a join
⌨️ (2:39:10) Uml diagrams for our current application
⌨️ (2:41:58) Associative vs aggregation
⌨️ (2:45:47) Introduction to compound queries
⌨️ (2:49:53) Fetching the list of all tracks
⌨️ (2:53:03) Using the delete statement to delete tracks
⌨️ (2:56:35) The problem with converting an integer to an integer
⌨️ (3:00:33) Deleting a track by deleting its Id

🎉 Thanks to our Champion and Sponsor supporters:
👾 Nattira Maneerat
👾 Heather Wcislo
👾 Serhiy Kalinets
👾 Erdeniz Unvan
👾 Justin Hual
👾 Agustín Kussrow
👾 Otis Morgan

--

Learn to code for free and get a developer job: https://www.freecodecamp.org

Read hundreds of articles on programming: https://freecodecamp.org/news

GPT-3 has taken the world by storm, and for good reason: It lets you generate text with just a few prompts (i.e. 'few-shot learning') or even with no prompts and just the instruction (i.e. 'zero-shot learning'). But there are scenarios where the model does not deliver satisfying results out of the box. In these cases we need to train the GPT-3 model with our own dataset to improve the performance. In this talk he explored how this can be done and we saw a real-life application of this.

This speech by Heiko Hotz was held on November 18 at Data Science Conference Europe 2022 in person in Belgrade.

Follow us on social media :
LinkedIn: Data Science Conference
https://www.linkedin.com/company/11184830
Instagram: @datasciconf
https://www.instagram.com/datasciconf/
Facebook page: @DataSciConference
https://www.facebook.com/DataSciConference/
Website: http://datasciconference.com/

Change the World Through Data

#datascience #ai #ml #artificialintelligence #bigdata #dataanalytics #gpt3

Learn how to use Vite, a build tool and development server designed specifically for modern web development. This course covers various topics, including creating a static server, using templates, integrating Tailwind, using environment variables, deploying to GitHub, Netlify, and Vercel, and configuring Vite. It also discusses the features of Vite, including hot module replacement and static asset handling.

✏️ Course created by @KhattakDev
YouTube: https://youtube.com/khattakdev
Twitter: https://twitter.com/khattakdev
Instagram: https://www.instagram.com/khattakdev

⭐️ Contents ⭐️
⌨️ (0:00:00) Start
⌨️ (0:00:22) Learning Objectives
⌨️ (0:01:29) Introduction
⌨️ (0:02:50) What makes it fast?
⌨️ (0:04:49) Creating Static Server
⌨️ (0:17:29) Using Templates
⌨️ (0:24:39) Tailwind Integration
⌨️ (0:32:30) Env Variables
⌨️ (0:40:58) Deployments
⌨️ (1:00:22) Configurations
⌨️ (1:30:35) Wrap Up

🎉 Thanks to our Champion and Sponsor supporters:
👾 Nattira Maneerat
👾 Heather Wcislo
👾 Serhiy Kalinets
👾 Erdeniz Unvan
👾 Justin Hual
👾 Agustín Kussrow
👾 Otis Morgan

--

Learn to code for free and get a developer job: https://www.freecodecamp.org

Read hundreds of articles on programming: https://freecodecamp.org/news

Learn about IBM watsonx→ https://ibm.biz/BdvkK8

LangChain became immensely popular when it was launched in 2022, but how can it impact your development and application of AI models, Large Language Models (LLM) in particular. In this video Martin Keen shares an overview of the features and uses of LangChain.

Get started for free on IBM Cloud → https://ibm.biz/sign-up-now
Subscribe to see more videos like this in the future → http://ibm.biz/subscribe-now

Get the guide to GAI, learn more → https://ibm.biz/BdKTbF
Learn more about the technology → https://ibm.biz/BdKTbX

Join Cedric Clyburn as he explores the differences and use cases of Retrieval Augmented Generation (RAG) and fine-tuning in enhancing large language models. This video covers the strengths, weaknesses, and common applications of both techniques, and provides insights on how to choose between them using machine learning and natural language processing principles

AI news moves fast. Sign up for a monthly newsletter for AI updates from IBM → https://www.ibm.com/account/re....g/us-en/signup?formi

#AI #LargeLanguageModels #FineTuning #RAG #ReinforcementLearning #MachineLearning #NaturalLanguageProcessing

Chat GPT 3 is an AI language Train Model that builds dialogue systems for customer engagement. It is the world’s first and only easy-to-use platform for training and deploying conversational AI systems.
#chatgpt #openai #chatgpt 3 #aicommunity #ai #newtechnologyideas #newtech

The KL divergence of distributions P and Q is a measure of how similar P and Q are.
However, the KL Divergence of P and Q is not the same as the KL Divergence of Q and P.
Why?
Learn the intuition behind this in this friendly video.

More about the KL Divergence formula:
https://www.youtube.com/watch?v=sjgZxuCm_8Q

An atom is mostly empty space, but empty space is mostly not empty. The reason it looks empty is because electrons and photons don't interact with the stuff that is there, quark and gluon field fluctuations.
It actually takes energy to clear out space and make a true 'empty' vacuum. This seems incredibly counter-intuitive but we can make an analogy to a permanent magnet. When at low energies, like at room temperature, there is a magnetic field around the magnet due to the alignment of all the magnetic moments of the atoms. But if you add some energy to it by heating it, the particles gain thermal energy, which above the Curie temperature makes their magnetic moments randomly oriented and hence destroying the magnetic field. So in this case energy is needed to clear out the field, just as in the quantum vacuum.

Special thanks to Professor Derek Leinweber, find out more about his research here: http://bit.ly/ZZTKFP

For a code implementation, check out this repo:
https://github.com/luisguiserr....ano/manning/tree/mas

Announcement: New Book by Luis Serrano! Grokking Machine Learning. bit.ly/grokkingML
40% discount code: serranoyt

An introduction to support vector machines (SVMs) that requires very little math (no calculus or linear algebra), only a visual mind.
This is the third of a series of three videos.
- Linear Regression: https://www.youtube.com/watch?v=wYPUhge9w5c
- Logistic Regression: https://www.youtube.com/watch?v=jbluHIgBmBo

0:00 Introduction
1:42 Classification goal: split data
3:14 Perceptron algorithm
6:00 Split data - separate lines
7:05 How to separate lines?
12:01 Expanding rate
18:19 Perceptron Error
19:26 SVM Classification Error
20:34 Margin Error
25:13 Challenge - Gradient Descent
27:25 Which line is better?
28:24 The C parameter
30:16 Series of 3 videos
30:30 Thank you!

🔥Professional Certificate Program in Generative AI and Machine Learning - IITG (India Only) - https://www.simplilearn.com/iitg-generative-ai-machine-learning-program?utm_campaign=labxPbQ88Zk&utm_medium=DescriptionFirstFold&utm_source=Youtube
🔥Purdue - Ai And Machine Learning Post Graduate Certificate Program - https://www.simplilearn.com/applied-ai-course?utm_campaign=labxPbQ88Zk&utm_medium=DescriptionFirstFold&utm_source=Youtube

In this video on openAI chatgpt 4o, we will be exploring the latest open ai model gpt4o, in this we will cover what is gpt40, how it different from the other gpt mode, chatgpt features demo like chitchatting with gpt, uploading image etc, at the end we will cover safety and limitations of gpt 4o.
GPT-4o is a big upgrade in OpenAI's smart tech. It's better at understanding and talking like a human than previous versions. It's super good at tasks like answering questions or writing stories. OpenAI keeps making it better. Now, it can even understand pictures! This makes it even more useful.

Below are the topics covered in how to use chatgpt 4o video:

00:00 Introduction To Open AI ChatGPT-4o
00:24 What Is Gpt 4o
02:21 How Gpt 4o is different from other models
02:44 Chatgpt 4o Demo Features
11:25 Safety and limitations of gpt 4o

✅Subscribe to our Channel to learn more about the top Technologies: https://bit.ly/2VT4WtH

#chatgpt4o # GPT4o #AI #ArtificialIntelligence #MachineLearning #2024 #Simplilearn

⏩ Check out More AI Videos By Simplilearn: https://www.youtube.com/watch?v=ad79nYk2keg&list=PLEiEAq2VkUULyr_ftxpHB6DumOq1Zz2hq&pp=gAQBiAQB

➡️ About Artificial Intelligence Engineer

This Artificial Intelligence Engineer course Created in partnership with IBM, this course introduces students to blended learning and prepares them to be AI and Data Science specialists. In Armonk, New York, IBM is a significant cognitive service and integrated cloud solution firm that provides many technology and consulting solutions.

IBM is a leader in AI and Machine Learning technology verticals for 2021. This AI masters course will prepare students for Artificial Intelligence and Data Analytics careers.

✅ Key Features

- Add the IBM Advantage to your Learning
- 25 Industry-relevant Projects and Integrated labs
- Immersive Learning Experience
- Simplilearn's JobAssist helps you get noticed by top hiring companies

✅ Skills Covered

- ChatGPT
- Flask
- Matplotlib
- django
- Python
- Numpy
- Pandas
- SciPy
- Keras
- OpenCV
- And Many More…

👉 Enroll Now: https://www.simplilearn.com/pgp-ai-machine-learning-certification-training-course?utm_campaign=17May24ChatGPT4o&utm_medium=Description&utm_source=youtube
🔥🔥 *Interested in Attending Live Classes? Call Us:* IN - 18002127688 / US - +18445327688

How to build your own Tiny Language Model from scratch - Luca Gilli - PyCon Italia 2025

Elevator Pitch:
Join us to discover Tiny Language Models: ultra-compact, resource-efficient AI tools that deliver impressive results. In this session, you’ll learn how to create your own model, from data preparation to training. These models combine innovation with efficiency, making AI more accessible than ever.

Learn more: https://2025.pycon.it/event/ho....w-to-build-your-own-

#MachineLearning #DeepLearning #NaturalLanguageProcessing

GPT-3 is super intelligent NLP deep learning model. In order to understand GPT-3 or later version, we should understand fundamental basic of it, and this video is covering the basic of GPT which covers before the transformer, and summary of transformer and the basic fundamental of GPT by looking at GPT-1.

Never get stuck without AI again. Run three Small Language Models (SLMs)—also called Local LLMs—TinyLlama, Gemma-3 and Phi-4-mini—completely offline; all fit in 4 GB or less and work on any laptop and older hardware.

────────────────────
🔧 Hardware & Software used
• Laptop Ryzen 5 4500U, 8GB RAM, Ollama (no GPU needed!)
• Phone iPhone 13 Pro with Mobile PocketPal AI (local GGUF)

────────────────────
🔗 Model resources
• ChatGPT global outage (news)
https://timesofindia.indiatime....s.com/etimes/trendin
• Phi-4-mini reasoning paper
https://www.microsoft.com/en-u....s/research/wp-conten
• TinyLlama 1.1 https://huggingface.co/TinyLlama/TinyLlama_v1.1
└ GGUF Q4_0 637 MB https://huggingface.co/TheBlok....e/TinyLlama-1.1B-Cha
• Gemma-3 https://huggingface.co/blog/gemma3
└ GGUF Q4_K_M 0.8 GB https://huggingface.co/Maziyar....Panahi/gemma-3-1b-it
• Phi-4-mini https://huggingface.co/microso....ft/Phi-4-mini-reason
└ GGUF Q4_K_M 2.5 GB https://huggingface.co/lmstudi....o-community/Phi-4-mi

────────────────────
🎬 More on local AI
• End of VRAM? https://youtu.be/M9ZphDPRP_w
• Is local AI image generation dying? https://youtu.be/ad7jBaNgIW8

🛠 Support the channel
Patreon https://www.patreon.com/NextTechAndAi

────────────────────
▼ Comment Poll
What would YOU use offline AI for?

#SmallLanguageModels #LocalLLM #OfflineLLM #LocalAI