Hello there! I'm Daniel, a 18yo idiot that builds AI.
AI-Atlas score
| Machine Learning | 67 |
|---|---|
| Computer Vision | 19 |
| Data Science | 19 |
| NLP & LLMs | 70 |
| AI Agents | 5 |
| MLOps | 12 |
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.
22 public repositories → 8 with AI work, evaluated. The other 14 (repositories without AI code, forks and empty repositories) are not scored and have no effect on the result. How repositories are selected
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.
42% of Atlas Score
LM-From-Scratch is an educational guide that walks through the process of designing and building a transformer-based language model from scratch using PyTorch. The repository includes documentation on architectural decisions and a code implementation of core components such as attention, feedforward networks, and training loops.
22% of Atlas Score
Open-SnowflakeCore is an educational implementation of a GPT-style transformer language model, consisting of two modules (G0 and G1) that provide configurable architectures with training loops and generation capabilities. The repository emphasizes learning through clear documentation and well-commented code.
16% of Atlas Score
The repository provides implementations of hybrid RWKV-Transformer models for BERT-style masked language modeling, language generation, and CLIP-style zero-shot image classification, including training and inference scripts.
10% of Atlas Score
The repository provides a low-rank factorized linear layer (LoRPtLinear) for memory-efficient pretraining, accompanied by extensive documentation describing its motivation, benchmark results, and ablation studies.
5% of Atlas Score
This repository hosts three research papers describing the i3 family of hybrid LLM architectures that combine RWKV/Mamba state-space mechanisms with Transformer attention. It provides PDF reports but no source code, models, or training scripts.
2% of Atlas Score
MuffinLM is a simple LSTM-based text generation model implemented in PyTorch, providing training and generation functionalities for educational purposes.
