26:13

Training LLMs on the entire connected web instead of individual sequences

15:48

Can AI change the way humanity communicates?

01:34:38

Flash-attention backward pass | Triton GPU Kernels 101 Lesson #10

01:33:26

Flash-attention forward pass | Triton GPU Kernels 101 Lesson #9

09:46

Compressing attention heads for BETTER performance?!?!

01:22:39

LayerNorm | Triton GPU Kernels 101 Lesson #8

01:20:14

Matmul | Triton GPU Kernels 101 Lesson #6

11:41

Dropout | Triton GPU Kernels 101 Lesson #7

01:05:13

Fused Softmax | Triton GPU Kernels 101 Lesson #5

41:07

Vector addition | Triton GPU Kernels 101 Lesson #4

18:22

How to use a cloud GPU | Triton GPU Kernels 101 Lesson #3

15:10

GPU Architecture Basics | Triton GPU Kernels 101 Lesson #2

15:40

Triton GPU Kernels 101: Syllabus day (Lesson #1)

01:04:34

Conversational Swarm Intelligence with Dr. Louis Rosenberg

13:14

Building models specifically for multi-GPU parallelization

07:21

GPU Kernels but with ✨pretty diagrams✨ to explain them

13:05

channel update

16:59

Has diffusion FINALLY been dethroned for image generation?!?!

33:33

How to make neural networks better at learning new things

26:53

Have we been doing LLM inference wrong the whole time?!?!

55:27

This month's HOTTEST🥵 new AI papers - February 2025

38:07

Train global AI models with Federated Learning using Flower (Tutorial / Interview)

20:09

How large groups can use AI to organize without leadership

13:35

MAJOR inference efficiency gain for diffusion models

17:30

Pre-train with patches for huge compute savings

15:08

Synthetic data that finally helps instead of hurts!!

23:15

Aligning AGI by converging its internal sense of self and other

14:18

Initializing large models with weights from smaller ones

30:53

Fine, I'll talk about the Meta papers y'all keep sending me

21:30

How diffusion modeling can revolutionize evolutionary algorithms

19:02

Let this method tune hyper-parameters for you!

39:21

Skimming hella new AI paper abstracts - January 2025

25:32

Mitigating frequency bias and anisotropy in LLM pre-training

49:09

2025 Yearly Theme and Productivity Scheme

22:08

Getting LLMs to look inward

16:21

Creating new tokens out of internal representations

25:17

Dark Representations: the error source SAEs cannot fit

25:09

Training GPTs on cellular automata

19:53

Training a GPT on massive amounts of fMRI data

23:00

Amateurs CAN do LLM research - but also, critiques are valuable

01:00:08

Michael Levin's newest paper - a commentary

27:52

Curating December's best new AI papers from ArXiv

16:25

Channel Update: what to expect over the next month+

20:59

Coding a single neuron w/ backprop in numpy | Deep-ML Code Challenges #25

24:57

Live-solving LeetCode #1679 - Max number of k-sum pairs

28:58

Live-solving LeetCode #334 - Increasing triplet subsequence

08:47

Live-solving LeetCode #1456 - Maximum number of vowels in a substring of given length

09:41

Live-solving LeetCode #283 - Move zeroes

13:51

Live-solving LeetCode #392 - Is subsequence

31:59

Live-solving LeetCode #11 - Container with most water

07:42

Live-solving LeetCode #1768 - Merge strings alternately

07:56

Live-solving LeetCode #1431 - Kids with the greatest number of candies

15:23

Live-solving LeetCode #1071 - Greatest common divisor of strings

10:28

Live-solving LeetCode #151 - Reverse words in a string

14:02

Live-solving LeetCode #605 - Can Place Flowers

09:12

Live-solving LeetCode #345 - Reverse Vowels of a String

01:05:35

Normalizing GPT on the unit-hypersphere (WITH CODE)

09:13

Theoretical physics of next token prediction in LLMs

51:50

Skimming hella AI paper abstracts - Nov 5, 2024

33:05

Every attention head explained