AI by Hand 2025 Calendar

12 Key Deep Learning Architectures

I made a 2025 calendar, featuring 12 key deep learning architectures I drew and calculated by hand ✍️

2025 is going to be a busy year. If you only have time to study one deep learning architecture each month, I would recommend the following:

1. January: Multi-Layer Perceptron (MLP)

2. February: Backpropagation

3. March: Autoencoder

4. April: Recurrent Neural Network (RNN)

5. May: Long-Short Term Memory (LSTM)

6. June: Generative Adversarial Network (GAN)

7. July: Residual Network (ResNet)

8. August: Transformer

9. September: Self-Attention

10. October: U-Net

11. November: Reinforcement Learning with Human Feedback (RLHF)

12. December: Contrastive Language-Image Pretraining (CLIP)

Some of you may notice CNN is missing. The reason is that I haven't managed to calculate it by hand yet. I hope I can do it in 2025.

I asked my wife which is her favorite. She said RNN. I asked why. She said "because it is most colorful!"

I asked my 7-year old the same question. He said CLIP. I asked why. He said "because it is most colorful!"

😄 Same reason but different answer!

Which is your favorite?

If you want a copy for yourself or gift it to someone, it is now available (USA only).