Unclaimed Atlas
Claim it with GitHub to join the leaderboard and see your rank, globally and in your country. Claiming also unlocks your full scoring analysis.
Checking your GitHub sign-in…
Your account, and you’d rather it wasn’t here? Request removal.
| Machine Learning | 74 |
|---|---|
| Computer Vision | 56 |
| Data Science | 53 |
| NLP & LLMs | 76 |
| AI Agents | 62 |
| MLOps | 22 |
Each node is one domain, scored only from evidence in these public repositories. Select one to see why.
Deep, repeated, well-engineered work in this domain.
Showing the top 30 of 37 evaluated projects by contribution.
Ordered by how much each project adds to the Atlas score, not by its own score. A project with evidence in several domains can rank above one with a slightly higher project score. Choose a domain to focus on the projects with meaningful evidence there.
+3.5 adoption
16% of Atlas Score
makemore is a single‑file PyTorch script that implements and trains various character‑level autoregressive language models (from bigrams to Transformers) for educational purposes.
+4.5 adoption
15% of Atlas Score
llama2.c provides a minimal educational implementation for training and inferencing the Llama 2 language model, featuring a PyTorch training pipeline, model export utilities, and a pure C inference engine for dependency-free execution.
+5 adoption
11% of Atlas Score
The repository provides a setup for an AI agent to autonomously conduct research by modifying a single-GPU LLM training script (train.py) and running experiments with a fixed 5-minute time budget, using a simplified nanochat-based model and custom optimizer.
+3.6 adoption
7% of Atlas Score
NeuralTalk is an educational implementation of image captioning models using multimodal recurrent neural networks (CNN+LSTM and CNN+RNN) built from scratch with Python and numpy.
+4.6 adoption
6% of Atlas Score
This repository contains a series of Jupyter notebooks that teach neural networks from scratch, implementing an autograd engine (micrograd) and character‑level language models up to a GPT‑like architecture.
+2.1 adoption
6% of Atlas Score
ResearchPooler is a collection of Python scripts that download, parse, and index NIPS conference papers from 2003-2010, enabling keyword and similarity-based search over the resulting bibliography.
