Daily Epoch
The library

Every paper that built modern AI

Start with a course

guided routes for newcomers
Free collection
18 landmark papers, free to read
A curated starter collection — the papers that built modern AI, each with a full reading page. Sign up to read them all.

Browse all 21 tracks

0114 guides
Foundations & Training Mechanics
Understand the nuts and bolts that make every model trainable.
1986Backpropagation1998LeNet2014Dropout+11 more →
not started
0210 guides
Sequence Models & Pre-Transformer NLP
See where modern NLP came from — and why attention was such a leap.
1997LSTM2014GRU (RNN Encoder–Decoder)2014Seq2Seq+7 more →
not started
039 guides
The Transformer Family
Understand the architecture behind essentially every modern AI model.
2017Attention Is All You Need2019Transformer-XL2021RoPE (RoFormer)+6 more →
not started
0410 guides
Encoder / Decoder Language Models
Learn the models that power search, classification and extraction (e.g. BERT/RoBERTa).
2018BERT2019RoBERTa2019ALBERT+7 more →
not started
0520 guides
GPT Line & Frontier LLMs
Understand how ChatGPT-class models are built and scaled.
2018GPT-12019GPT-22020GPT-3+17 more →
not started
0619 guides
Alignment, Instruction & Reasoning
Learn how a raw model becomes helpful, safe, and able to reason.
2022InstructGPT2017RLHF (Christiano)2017PPO+16 more →
not started
0717 guides
Scaling, Efficiency & Systems
Make models cheap enough to train and fast enough to ship.
2020Kaplan Scaling Laws2022Chinchilla2021Switch Transformer+14 more →
not started
0813 guides
CNN Backbones
Understand the workhorses of image recognition.
2012AlexNet2014VGG2014GoogLeNet / Inception+10 more →
not started
095 guides
Vision Transformers
See how Transformers took over computer vision.
2020ViT2021Swin Transformer2021DeiT+2 more →
not started
1016 guides
Object Detection & Segmentation
Build systems that find and outline objects, not just label them.
2015Faster R-CNN2017Mask R-CNN2016YOLOv1+13 more →
not started
1113 guides
Self-Supervised & Representation Learning
Train powerful models without labeled data.
2020SimCLR2019MoCo2018CPC+10 more →
not started
1211 guides
Generative — GANs
Understand the models that first made convincing synthetic imagery.
2014GANs2016DCGAN2017WGAN+8 more →
not started
135 guides
Generative — VAE & Autoregressive
Learn the probabilistic side of generation behind today's tokenizers.
2013VAE2017VQ-VAE2019VQ-VAE-2+2 more →
not started
1415 guides
Generative — Diffusion & Flow
Understand the engine behind Midjourney, Stable Diffusion & Sora.
2020DDPM2020DDIM2021Diffusion Beats GANs+12 more →
not started
154 guides
3D & Neural Rendering
Explore how AI captures and renders 3D scenes.
2020NeRF2022Instant-NGP20233D Gaussian Splatting+1 more →
not started
1614 guides
Multimodal & Vision-Language
Learn how models that see AND read (like GPT-4V) actually work.
2021CLIP2023SigLIP2021ALIGN+11 more →
not started
175 guides
Speech & Audio
Understand the models behind voice assistants and transcription.
2022Whisper2020Wav2Vec 2.02021HuBERT+2 more →
not started
1810 guides
Reinforcement Learning
Learn how agents master games and control by trial and error.
2013DQN (Playing Atari)2016A3C2015DDPG+7 more →
not started
1910 guides
World Models & Agents
See the frontier: agents that plan, use tools, and imagine.
2018World Models2023DreamerV32024DINO-WM+7 more →
not started
203 guides
Graph Neural Networks
Learn to model relational data — molecules, networks and knowledge graphs.
2017GraphSAGE2018Graph Attention Networks2017GCNOpen track →
not started
212 guides
Interpretability & Robustness
Understand why a model predicts what it does — and how it can be fooled.
2017SHAP2018Adversarial Robustness (PGD)Open track →
not started