Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

KL divergence is a fundamental concept used in machine learning, from optimization a neural net to RL training of LLMs. but here is what it actually means:

ℏεsam

80,475 subscribers

20,381 views • 6 months ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Our course recommendation of the day is “Post-training of LLMs, ” where you’ll learn how to customize pre-trained language models using Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL). You'll learn when to use each method, how to curate training data, and implement them in code to shape model behavior effectively. Enroll here:

Our course recommendation of the day is “Post-training of LLMs, ” where you’ll learn how to customize pre-trained language models using Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL). You'll learn when to use each method, how to curate training data, and implement them in code to shape model behavior effectively. Enroll here:

DeepLearning.AI

29,369 views • 10 months ago

David Deutsch says all computation from humans to LLMs shares the same fundamental power: the universal Turing machine The difference is only in programs and outputs. But AGI is different; we don't even understand it in principle. "so the true sign would be a theory, not a machine"

David Deutsch says all computation from humans to LLMs shares the same fundamental power: the universal Turing machine The difference is only in programs and outputs. But AGI is different; we don't even understand it in principle. "so the true sign would be a theory, not a machine"

Haider.

11,720 views • 11 months ago

Does GPT understand the world? Here is what Ilya Sutskever, co-founder of OpenAI, says during a discussion with Jensen Huang, CEO of Nvidia: (1) When we train a large neural network to accurately predict the next word in lots of different texts from the internet, the AI is learning a world model. (2) On the surface, it may look like learning correlations in text, but it turns out that to 'just learn' statistical correlations in text, to compress information really well, what the neural network learns is some representation of the process that produced the text. (3) This text is a projection of the world...what the neural network is learning is aspects of the world, of people, of the human conditions, their hopes, dreams, motivations, their interactions...the situations we are in. The neural network learns a compressed, abstract, usable representation." Do you think learning representations = understanding? Are large language models simply stochastic parrots, or are they much more?

Does GPT understand the world? Here is what Ilya Sutskever, co-founder of OpenAI, says during a discussion with Jensen Huang, CEO of Nvidia: (1) When we train a large neural network to accurately predict the next word in lots of different texts from the internet, the AI is learning a world model. (2) On the surface, it may look like learning correlations in text, but it turns out that to 'just learn' statistical correlations in text, to compress information really well, what the neural network learns is some representation of the process that produced the text. (3) This text is a projection of the world...what the neural network is learning is aspects of the world, of people, of the human conditions, their hopes, dreams, motivations, their interactions...the situations we are in. The neural network learns a compressed, abstract, usable representation." Do you think learning representations = understanding? Are large language models simply stochastic parrots, or are they much more?

Alex Ker 🔭

1,367,097 views • 2 years ago

2/ Let’s start with the concept of #Nakba - how it started, what it means and how it evolved from the concept of “Arab humiliation” to “the massacre of Arabs by Zionists” (This is from a conversation in Hebrew, the translation is AI, not perfect but good enough) >>

2/ Let’s start with the concept of #Nakba - how it started, what it means and how it evolved from the concept of “Arab humiliation” to “the massacre of Arabs by Zionists” (This is from a conversation in Hebrew, the translation is AI, not perfect but good enough) >>

Gal G., Adv 🇮🇱🌵

29,550 views • 1 year ago

Data preparation! It's crucial for machine learning, and we all hate it. Tools and techniques to reduce this burden? A quick summary of 10 years of R&D on this, from cheap tricks to LLMs and graph neural networks 1/9

Data preparation! It's crucial for machine learning, and we all hate it. Tools and techniques to reduce this burden? A quick summary of 10 years of R&D on this, from cheap tricks to LLMs and graph neural networks 1/9

Gael Varoquaux 🦋

13,757 views • 1 year ago

📁 Geoffrey Hinton, a pioneer of deep learning, says LLMs do understand what they say. He explains that understanding language is not translating it into an internal symbolic code. In neural networks, words act like Lego blocks that combine to form meaning. It is not translation, it is structure. That is what real language understanding looks like.

📁 Geoffrey Hinton, a pioneer of deep learning, says LLMs do understand what they say. He explains that understanding language is not translating it into an internal symbolic code. In neural networks, words act like Lego blocks that combine to form meaning. It is not translation, it is structure. That is what real language understanding looks like.

Jon Hernandez

77,752 views • 6 months ago

AI is changing everything, but how does it actually work? We break down machine learning, training, inference, and what happens behind the scenes when you use AI in the latest episode of #TheBringUp, featuring a hands-on model demo. 🎥 Watch now:

AI is changing everything, but how does it actually work? We break down machine learning, training, inference, and what happens behind the scenes when you use AI in the latest episode of #TheBringUp, featuring a hands-on model demo. 🎥 Watch now:

AMD

23,492 views • 1 month ago

Did you know that NaN exists not only to troll JavaScript programmers - but it is actually used for NaN Boxing, which is a very cool optimization technique!

Did you know that NaN exists not only to troll JavaScript programmers - but it is actually used for NaN Boxing, which is a very cool optimization technique!

Valentin Ignatev

29,959 views • 5 months ago

Φ-SO : Physical Symbolic Optimization - Learning Physics from Data 🧠 The Physical Symbolic Optimization package uses deep reinforcement learning to discover physical laws from data. Here is Φ-SO discovering the analytical expression of a damped harmonic oscillator.

Φ-SO : Physical Symbolic Optimization - Learning Physics from Data 🧠 The Physical Symbolic Optimization package uses deep reinforcement learning to discover physical laws from data. Here is Φ-SO discovering the analytical expression of a damped harmonic oscillator.

Jousef Murad

433,199 views • 2 years ago

Net Zero is one of the biggest goals in climate today. But what does it actually take to get there? #dkarbon

Net Zero is one of the biggest goals in climate today. But what does it actually take to get there? #dkarbon

dkarbon

34,909 views • 5 days ago

Before the throne. Before the legend. Just a boy learning what it means to be great. A Knight of the Seven Kingdoms is here. 🕯️ #AKnightOfTheSevenKingdoms #GameOfThrones #GRRM

Before the throne. Before the legend. Just a boy learning what it means to be great. A Knight of the Seven Kingdoms is here. 🕯️ #AKnightOfTheSevenKingdoms #GameOfThrones #GRRM

Javed

52,305 views • 5 months ago

.Eric Jang explains the fundamental intuition behind AlphaGo: experienced Go players almost always end games early. That means they have the ability to glance at a board and instantly know whether it's hopeless or if they should keep playing. Without this ability, the only way to know if a move is good is to play out every possible game from that point - completely intractable. But the ability is clearly trainable in humans. So it should be possible to get a neural net to learn it. And if you have a good enough neural net, you can use it to ignore moves that will definitely make you lose. Suddenly search becomes tractable.

.Eric Jang explains the fundamental intuition behind AlphaGo: experienced Go players almost always end games early. That means they have the ability to glance at a board and instantly know whether it's hopeless or if they should keep playing. Without this ability, the only way to know if a move is good is to play out every possible game from that point - completely intractable. But the ability is clearly trainable in humans. So it should be possible to get a neural net to learn it. And if you have a good enough neural net, you can use it to ignore moves that will definitely make you lose. Suddenly search becomes tractable.

Dwarkesh Patel

49,692 views • 2 months ago

if you're struggling on where to start learning ML, here’s a playlist of 30 youtube videos to learn machine learning fundamentals from scratch "Machine Learning: Teach by Doing" is a solid choice to learn both theory and code. (1) Introduction to Machine Learning Teach by Doing: (2) What is Machine Learning? History of Machine Learning: (3) Types of ML Models: (4) 6 steps of any ML project: (5) Install Python and VSCode and run your first code: (6) Linear Classifiers Part 1: (7) Linear Classifiers Part 2: (8) Jupyter Notebook, Numpy and Scikit-Learn: (9) Running the Random Linear Classifier Algorithm in Python: (10) The oldest ML model - Perceptron: (11) Coding the Perceptron: (12) Perceptron Convergence Theorem: (13) Magic of features in Machine Learning: (14) One hot encoding: (15) Logistic Regression Part 1: (16) Cross Entropy Loss: (17) How gradient descent works: (18) Logistic Regression from scratch in Python: (19) Introduction to Regularization: (20) Implementing Regularization in Python: (21) Linear Regression Introduction: (22) Ordinary Least Squares step by step implementation: (23) Ridge regression fundamentals and intuition: (24) Regression recap for interviews: (25) Neural network architecture in 30 minutes: (26) Backpropagation intuition: (27) Neural network activation functions: (28) Momentum in gradient descent: (29) Hands on neural network training in Python: (30) Introduction to Convolutional Neural Networks (CNNs):

if you're struggling on where to start learning ML, here’s a playlist of 30 youtube videos to learn machine learning fundamentals from scratch "Machine Learning: Teach by Doing" is a solid choice to learn both theory and code. (1) Introduction to Machine Learning Teach by Doing: (2) What is Machine Learning? History of Machine Learning: (3) Types of ML Models: (4) 6 steps of any ML project: (5) Install Python and VSCode and run your first code: (6) Linear Classifiers Part 1: (7) Linear Classifiers Part 2: (8) Jupyter Notebook, Numpy and Scikit-Learn: (9) Running the Random Linear Classifier Algorithm in Python: (10) The oldest ML model - Perceptron: (11) Coding the Perceptron: (12) Perceptron Convergence Theorem: (13) Magic of features in Machine Learning: (14) One hot encoding: (15) Logistic Regression Part 1: (16) Cross Entropy Loss: (17) How gradient descent works: (18) Logistic Regression from scratch in Python: (19) Introduction to Regularization: (20) Implementing Regularization in Python: (21) Linear Regression Introduction: (22) Ordinary Least Squares step by step implementation: (23) Ridge regression fundamentals and intuition: (24) Regression recap for interviews: (25) Neural network architecture in 30 minutes: (26) Backpropagation intuition: (27) Neural network activation functions: (28) Momentum in gradient descent: (29) Hands on neural network training in Python: (30) Introduction to Convolutional Neural Networks (CNNs):

ℏεsam

109,292 views • 1 year ago

a playlist of 30 youtube videos to learn machine learning fundamentals from scratch if you're struggling on where to start learning ML, this list goes this "Machine Learning: Teach by Doing" is a solid choice to learn both theory and code. (1) Introduction to Machine Learning Teach by Doing: (2) What is Machine Learning? History of Machine Learning: (3) Types of ML Models: (4) 6 steps of any ML project: (5) Install Python and VSCode and run your first code: (6) Linear Classifiers Part 1: (7) Linear Classifiers Part 2: (8) Jupyter Notebook, Numpy and Scikit-Learn: (9) Running the Random Linear Classifier Algorithm in Python: (10) The oldest ML model - Perceptron: (11) Coding the Perceptron: (12) Perceptron Convergence Theorem: (13) Magic of features in Machine Learning: (14) One hot encoding: (15) Logistic Regression Part 1: (16) Cross Entropy Loss: (17) How gradient descent works: (18) Logistic Regression from scratch in Python: (19) Introduction to Regularization: (20) Implementing Regularization in Python: (21) Linear Regression Introduction: (22) Ordinary Least Squares step by step implementation: (23) Ridge regression fundamentals and intuition: (24) Regression recap for interviews: (25) Neural network architecture in 30 minutes: (26) Backpropagation intuition: (27) Neural network activation functions: (28) Momentum in gradient descent: (29) Hands on neural network training in Python: (30) Introduction to Convolutional Neural Networks (CNNs):

a playlist of 30 youtube videos to learn machine learning fundamentals from scratch if you're struggling on where to start learning ML, this list goes this "Machine Learning: Teach by Doing" is a solid choice to learn both theory and code. (1) Introduction to Machine Learning Teach by Doing: (2) What is Machine Learning? History of Machine Learning: (3) Types of ML Models: (4) 6 steps of any ML project: (5) Install Python and VSCode and run your first code: (6) Linear Classifiers Part 1: (7) Linear Classifiers Part 2: (8) Jupyter Notebook, Numpy and Scikit-Learn: (9) Running the Random Linear Classifier Algorithm in Python: (10) The oldest ML model - Perceptron: (11) Coding the Perceptron: (12) Perceptron Convergence Theorem: (13) Magic of features in Machine Learning: (14) One hot encoding: (15) Logistic Regression Part 1: (16) Cross Entropy Loss: (17) How gradient descent works: (18) Logistic Regression from scratch in Python: (19) Introduction to Regularization: (20) Implementing Regularization in Python: (21) Linear Regression Introduction: (22) Ordinary Least Squares step by step implementation: (23) Ridge regression fundamentals and intuition: (24) Regression recap for interviews: (25) Neural network architecture in 30 minutes: (26) Backpropagation intuition: (27) Neural network activation functions: (28) Momentum in gradient descent: (29) Hands on neural network training in Python: (30) Introduction to Convolutional Neural Networks (CNNs):

ℏεsam

117,570 views • 1 year ago

RL is a powerful mechanism for training company-specific models on their unique work and data. This is what we do at Applied Compute. A key challenge is how to make RL efficient, because we need runs to be fast (delivered in days), cheap (scalable unit economics), and predictable (not just fast, but reliably fast). Here are some takeaways: • Synchronous RL is wasteful with time and compute. • Asynchronous RL is more efficient but introduces staleness, which causes learning instabilities. • Modeling and simulations can help analytically solve for what configuration leads to optimal efficiency. This allows us to rapidly prototype training configurations, without burning expensive compute cycles on trial runs. Two of our co-founders, Rhythm Garg and Linden Li, discussed some of this research at AI Engineer recently, with a focus on the following subproblem: what is the highest throughput way to do RL given a maximum staleness and compute budget?

RL is a powerful mechanism for training company-specific models on their unique work and data. This is what we do at Applied Compute. A key challenge is how to make RL efficient, because we need runs to be fast (delivered in days), cheap (scalable unit economics), and predictable (not just fast, but reliably fast). Here are some takeaways: • Synchronous RL is wasteful with time and compute. • Asynchronous RL is more efficient but introduces staleness, which causes learning instabilities. • Modeling and simulations can help analytically solve for what configuration leads to optimal efficiency. This allows us to rapidly prototype training configurations, without burning expensive compute cycles on trial runs. Two of our co-founders, Rhythm Garg and Linden Li, discussed some of this research at AI Engineer recently, with a focus on the following subproblem: what is the highest throughput way to do RL given a maximum staleness and compute budget?

Applied Compute

45,555 views • 8 months ago

I like to think of evals as something active, not passive -- it's a North Star that steers LLMs toward higher intelligence. Evals should drive your RL/SFT/post-training decisions. Internal evals at frontier labs make a huge difference -- and you can see it in how models behave differently (GPT seems better at one-shot tasks, Claude at multi-turn). If you want to learn more about building evals that actually improve your model in post-training, check out our AMD x DeeplearningAI course "Fine-tuning & RL for LLMs: Intro to Post-training" (content is free):

I like to think of evals as something active, not passive -- it's a North Star that steers LLMs toward higher intelligence. Evals should drive your RL/SFT/post-training decisions. Internal evals at frontier labs make a huge difference -- and you can see it in how models behave differently (GPT seems better at one-shot tasks, Claude at multi-turn). If you want to learn more about building evals that actually improve your model in post-training, check out our AMD x DeeplearningAI course "Fine-tuning & RL for LLMs: Intro to Post-training" (content is free):

Sharon Zhou

16,836 views • 5 months ago

We asked Sholto Douglas from Anthropic about the costs of RL (Reinforcement Learning) runs. "In Dario Amodei's essay, he said that RL runs cost only $1M back in December." "RL is a more naively parallelizable and scalable than pre-training." "With pre-training, you need everything in one big data center ideally. For RL, in theory, you could scale all over the world."

We asked Sholto Douglas from Anthropic about the costs of RL (Reinforcement Learning) runs. "In Dario Amodei's essay, he said that RL runs cost only $1M back in December." "RL is a more naively parallelizable and scalable than pre-training." "With pre-training, you need everything in one big data center ideally. For RL, in theory, you could scale all over the world."

TBPN

76,696 views • 1 year ago

A culture of divination has grown all around us - from machine learning, to prediction markets, and forecasting of all kinds, not to mention psychics and astrology. But have you ever wondered what, exactly, is in a prediction? TED talk: #PROPHECY

A culture of divination has grown all around us - from machine learning, to prediction markets, and forecasting of all kinds, not to mention psychics and astrology. But have you ever wondered what, exactly, is in a prediction? TED talk: #PROPHECY

Carissa Véliz

30,068 views • 3 months ago

New Course: Post-training of LLMs Learn to post-train and customize an LLM in this short course, taught by Banghua Zhu, Assistant Professor at the University of Washington University of Washington, and co-founder of @NexusflowX. Training an LLM to follow instructions or answer questions has two key stages: pre-training and post-training. In pre-training, it learns to predict the next word or token from large amounts of unlabeled text. In post-training, it learns useful behaviors such as following instructions, tool use, and reasoning. Post-training transforms a general-purpose token predictor—trained on trillions of unlabeled text tokens—into an assistant that follows instructions and performs specific tasks. Because it is much cheaper than pre-training, it is practical for many more teams to incorporate post-training methods into their workflows than pre-training. In this course, you’ll learn three common post-training methods—Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL)—and how to use each one effectively. With SFT, you train the model on pairs of input and ideal output responses. With DPO, you provide both a preferred (chosen) and a less preferred (rejected) response and train the model to favor the preferred output. With RL, the model generates an output, receives a reward score based on human or automated feedback, and updates the model to improve performance. You’ll learn the basic concepts, common use cases, and principles for curating high-quality data for effective training. Through hands-on labs, you’ll download a pre-trained model from Hugging Face and post-train it using SFT, DPO, and RL to see how each technique shapes model behavior. In detail, you’ll: - Understand what post-training is, when to use it, and how it differs from pre-training. - Build an SFT pipeline to turn a base model into an instruct model. - Explore how DPO reshapes behavior by minimizing contrastive loss—penalizing poor responses and reinforcing preferred ones. - Implement a DPO pipeline to change the identity of a chat assistant. - Learn online RL methods such as Proximal Policy Optimization (PPO) and Group Relative Policy Optimization (GRPO), and how to design reward functions. - Train a model with GRPO to improve its math capabilities using a verifiable reward. Post-training is one of the most rapidly developing areas of LLM training. Whether you’re building a high-accuracy context-specific assistant, fine-tuning a model's tone, or improving task-specific accuracy, this course will give you experience with the most important techniques shaping how LLMs are post-trained today. Please sign up here:

New Course: Post-training of LLMs Learn to post-train and customize an LLM in this short course, taught by Banghua Zhu, Assistant Professor at the University of Washington University of Washington, and co-founder of @NexusflowX. Training an LLM to follow instructions or answer questions has two key stages: pre-training and post-training. In pre-training, it learns to predict the next word or token from large amounts of unlabeled text. In post-training, it learns useful behaviors such as following instructions, tool use, and reasoning. Post-training transforms a general-purpose token predictor—trained on trillions of unlabeled text tokens—into an assistant that follows instructions and performs specific tasks. Because it is much cheaper than pre-training, it is practical for many more teams to incorporate post-training methods into their workflows than pre-training. In this course, you’ll learn three common post-training methods—Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Online Reinforcement Learning (RL)—and how to use each one effectively. With SFT, you train the model on pairs of input and ideal output responses. With DPO, you provide both a preferred (chosen) and a less preferred (rejected) response and train the model to favor the preferred output. With RL, the model generates an output, receives a reward score based on human or automated feedback, and updates the model to improve performance. You’ll learn the basic concepts, common use cases, and principles for curating high-quality data for effective training. Through hands-on labs, you’ll download a pre-trained model from Hugging Face and post-train it using SFT, DPO, and RL to see how each technique shapes model behavior. In detail, you’ll: - Understand what post-training is, when to use it, and how it differs from pre-training. - Build an SFT pipeline to turn a base model into an instruct model. - Explore how DPO reshapes behavior by minimizing contrastive loss—penalizing poor responses and reinforcing preferred ones. - Implement a DPO pipeline to change the identity of a chat assistant. - Learn online RL methods such as Proximal Policy Optimization (PPO) and Group Relative Policy Optimization (GRPO), and how to design reward functions. - Train a model with GRPO to improve its math capabilities using a verifiable reward. Post-training is one of the most rapidly developing areas of LLM training. Whether you’re building a high-accuracy context-specific assistant, fine-tuning a model's tone, or improving task-specific accuracy, this course will give you experience with the most important techniques shaping how LLMs are post-trained today. Please sign up here:

Andrew Ng

125,146 views • 1 year ago