PaperarXiv · 2017
Attention Is All You Need
Vaswani et al.
Why it's here The paper that introduced the transformer. Section 3 is more readable than its reputation suggests.
HeroFree
> open ./resources
The good stuff.
Papers, courses, tools and datasets we actually used and would recommend to a friend. Every pick comes with a note on why it's here.
Starter packs
Starter pack
LLM basics
12 picks · about 14 hours
Starter pack
Building agents
9 picks · about 10 hours
Starter pack
Math for ML
10 picks · about 40 hours
The library
PaperarXiv · 2020
Language Models are Few-Shot Learners
Brown et al. (GPT-3)
Why it's here Where "just show it a few examples in the prompt" became a thing. Skim the results, read the discussion.
LegendFree
PaperarXiv · 2021
LoRA: Low-Rank Adaptation of Large Language Models
Hu et al.
Why it's here The reason you can fine-tune on a laptop. The core idea fits on one page.
HeroFree
PaperarXiv · 2020
Denoising Diffusion Probabilistic Models
Ho, Jain, Abbeel
Why it's here The paper behind modern image generators. Pair it with our issue #009 for the intuition first.
LegendFree
PaperarXiv · 2020
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Lewis et al.
Why it's here The paper that named RAG. Useful for seeing how simple the original idea was.
HeroFree
CourseYouTube
Neural Networks: Zero to Hero
Andrej Karpathy
Why it's here Builds everything from a scalar autograd engine up to GPT, in code, on camera. The best free course out there.
RookieFree
Coursefast.ai
Practical Deep Learning for Coders
Jeremy Howard
Why it's here Top-down teaching: train a working model in lesson one, learn the theory after. Great if you learn by doing.
RookieFree
Course3Blue1Brown
Neural networks (video series)
Grant Sanderson
Why it's here The visuals that make backprop and attention click. Watch before reading any paper.
RookieFree
CourseHugging Face
The LLM Course
Hugging Face
Why it's here The practical side: loading models, tokenizers and fine-tuning with the libraries everyone actually uses.
RookieFree
BookMIT Press · 2016
Deep Learning
Goodfellow, Bengio, Courville
Why it's here The reference textbook. Older than transformers, but the foundations haven't changed. Free to read online.
HeroFree
BookCambridge · 2020
Mathematics for Machine Learning
Deisenroth, Faisal, Ong
Why it's here Exactly the linear algebra, calculus and probability you need, and nothing else. Free PDF.
HeroFree
ToolOpen source
Transformers library
Hugging Face
Why it's here Load almost any open model in three lines. The default starting point for hands-on work.
RookieFree
ToolOpen source
llama.cpp
ggml-org
Why it's here Run quantized models on a laptop CPU or GPU. Reading its source is a crash course in inference tricks.
HeroFree
ToolSaaS
Weights & Biases
Weights & Biases
Why it's here Experiment tracking that makes "which run was the good one?" answerable. Free tier for personal projects.
HeroFree tier
DatasetHugging Face · 2024
FineWeb
Hugging Face
Why it's here A huge, cleaned web-text dataset for pretraining, with a write-up explaining every filtering decision.
LegendFree
DatasetBenchmark · 2020
MMLU
Hendrycks et al.
Why it's here Multiple-choice questions across 57 subjects. Read a few to see what "benchmark score" really measures.
HeroFree
ModelMistral AI · 2023
Mistral 7B
Mistral AI · Apache 2.0
Why it's here A small open model with a permissive license. A great base for your first fine-tuning experiments.
HeroFree
ModelMeta · 2024
Llama 3.1 8B
Meta · Llama license
Why it's here One of the most widely used open-weight models, so almost every tool and tutorial supports it.
HeroFree
No resources match. The AI is as confused as you are. Try removing a filter.