Every Confusing Thing About Neural Networks Explained Slowly (For Sleep)
Cosmo Explains
36,926 views
Now streaming on Spotify
https://open.spotify.com/show/033FVSz...
Take a slow journey through the fascinating history of neural networks from the first mathematical models of neurons and the perceptron to backpropagation, deep learning, AlexNet, ResNet, Transformers, and today’s large language models.
Along the way, we explore how machines actually learn, why neural networks disappeared and returned, how attention changed AI forever, and why even today we still don’t fully understand what happens inside these powerful systems. A calm, detailed look at one of the most important ideas in modern computing.
Resources
Nature — Learning Representations by Back-Propagating Errors
https://www.nature.com/articles/323533a0
Nature — Deep Learning
https://www.nature.com/articles/natur...
NeurIPS — Attention Is All You Need
https://papers.nips.cc/paper/7181-att...
GPT-3 — Language Models Are Few-Shot Learners
https://arxiv.org/abs/2005.14165
5 Comments
Constellation ⚡ Energy Corporation (CEG).
Every Confusing Thing About Neural Networks Explained Slowly (For Sleep)
Cosmo Explains
36,926 views
Now streaming on Spotify
https://open.spotify.com/show/033FVSz...
Take a slow journey through the fascinating history of neural networks from the first mathematical models of neurons and the perceptron to backpropagation, deep learning, AlexNet, ResNet, Transformers, and today’s large language models.
Along the way, we explore how machines actually learn, why neural networks disappeared and returned, how attention changed AI forever, and why even today we still don’t fully understand what happens inside these powerful systems. A calm, detailed look at one of the most important ideas in modern computing.
Resources
Nature — Learning Representations by Back-Propagating Errors
https://www.nature.com/articles/323533a0
Nature — Deep Learning
https://www.nature.com/articles/natur...
NeurIPS — Attention Is All You Need
https://papers.nips.cc/paper/7181-att...
GPT-3 — Language Models Are Few-Shot Learners
https://arxiv.org/abs/2005.14165