The Papers Bootcamp
For engineers who want to read the canonical deep-learning papers in the right order.
The papers that built modern AI, sequenced so each one earns the next — from word2vec and the Transformer through scaling laws, RLHF, chain-of-thought, RAG, LoRA and MoE, out to vision and diffusion. Every entry is a plain-English breakdown, not the raw PDF.
- Read the canonical papers in a coherent, building order
- Connect each paper to the capability it unlocked
- Trace the line from word vectors to reasoning models
- Follow the field beyond language into vision and diffusion
Word2Vec
Distributed word representations — the seed.
Seq2Seq
Encoder–decoder sequence learning.
Attention Is All You Need
The Transformer.
BERT
Bidirectional pretraining — the encoder line.
GPT-3
Few-shot learning at scale — the decoder line.
Scaling Laws
Why scale works, quantified.
Chinchilla
Compute-optimal training — data matters as much as size.
InstructGPT
Alignment via human feedback — RLHF.
Chain-of-Thought Prompting
Let models think step by step.
Self-Consistency
Sample many chains of thought, then vote.
Retrieval-Augmented Generation
Ground generation in retrieval.
LoRA
Fine-tune cheaply — low-rank adapters.
Direct Preference Optimization
Alignment without a separate reward model.
Mixture of Experts
Scale parameters without scaling compute.
FlashAttention
Make attention IO-aware and fast.
ReAct
Interleave reasoning and acting — agents.
Toolformer
Teach a model to call tools.
DeepSeek-R1
Reasoning that emerges from reinforcement learning.
Vision Transformer (ViT)
Transformers leave language — ViT.
CLIP
Connect images and text.
Diffusion Models
The generative engine behind modern image models.
Whisper
Weakly-supervised speech recognition at scale.