Code for my Deep Learning Course at University of Latvia
- Intro to Deep Learning: https://youtu.be/2K2RMy-5ZsM
- Bigram Model: https://youtu.be/Kbc-8iEe6Js (Original: https://youtu.be/kCc8FmEb1nY )
- Imitation learning (Data is King): https://youtu.be/8t4sxp7NVjg
- Numerical Optimisation: https://youtu.be/u-Sn0f2XHnI
- Architectures of Neural Networks: https://youtu.be/y6emEiI1lZw
- Intro to ChatGPT: https://youtu.be/HY-1RoVFftc (Course on prompt engineering ; GPT3 explained )
- Intro to Robotics: https://youtu.be/Bi9QtQb3vrA (Image segmentation example: https://github.com/paulispaulis/CLIC-semseg ; KarpathyAtTesla )
- From Bigram to ChatGPT: https://youtu.be/lZM36LLXWbc (Tokenization: https://youtu.be/zduSFxRajkE ; Attention: https://youtu.be/eMlx5fFNoYc ; What Transformers can learn: https://arxiv.org/abs/2310.16028)
- What is Language? https://youtu.be/aJ-3botLBM0 (Diversity of modern languages implies they emerged AFTER HomoSapiens migration out of Africa https://youtu.be/rQv7NBGsldk ; Impossible languages)
- Compression by LLM: https://youtu.be/r-fSI7rSjYI (Original: https://youtu.be/zjkBMFhNj_g )
- Evolutionary Learning: https://youtu.be/4OTuG0WRjaU (Demo https://youtu.be/GOFws_hhZs8 )
- Neural Network Basics: https://youtu.be/5KNE_zvnNeA
- Neural Network training (Backpropagation): https://youtu.be/GSZnx7VK3Zg (DetailedExplanation: https://youtu.be/VMj-3S1tku0 )
- PyTorch library: https://youtu.be/NlzC9zTXTHM
- Innovation: https://youtu.be/1Vhh_dYOu0I (Mystery2: Agentive evolutionary learning needs no training data)
- Technological Evolution https://youtu.be/wIJnvCqKPXM (Mystery1: Incremental 1-shot memory for continuous learning)
- Egocentric Viewpoint https://youtu.be/t3-InC9VEko (Mystery3: Concept creation)
In video description (press "...more" to reveal it all), at the lower-left corner there is a button "Show transcrpt". Copy/paste transcript to your favourite editor (remove timestamps with "Toggle timestamps" under symbol ⋮ in the top-right corner).
Videos were subtitled and voiced using SELMA UC0 OpenSource software available at: https://github.com/SELMA-project/UC0-OpenSource