I built neetcode-gpt, a Python GPT project on my GitHub, from my NeetCode ML course submissions. The README presents the repository as coursework progressing toward a working GPT, and it is listed on my github profile.
My neetcode-gpt repository dates are Started 2026-08, last updated 2026-08.
Every file in this project is code I wrote and submitted while completing the NeetCode ML course.
The problems progressively build from gradient descent fundamentals all the way to a working GPT.
The README organizes the code into the model, data pipeline, and foundations areas. The model area covers attention and Transformer/GPT architecture; data covers preparation and tokenization; foundations contains neural-network primitives built from scratch.
- The README lists these model components: Self-attention head; multi-headed attention; Transformer block; GPT model; layer normalization; batch normalization; RMS normalization; word embeddings; positional encoding; KV-Cache for fast inference; grouped query attention.
- The README lists these data components: BPE tokenizer; character-level vocabulary; batched training data loader; GPT dataset preparation; NLP preprocessing; tokenization edge cases.
- The README lists these foundations files: neuron.py, backprop.py, mlp.py, activations.py, loss.py, training_loop.py, dead_relu_detector.py.
My repository's top-level scripts are train.py GPT training loop; generate.py Text generation.
- The README gives this course topic sequence: Math Foundations (gradient descent, activations, loss functions); Neural Networks from scratch (neuron, backprop, MLP); PyTorch fundamentals; NLP pipeline (embeddings, tokenization, attention); Transformer architecture; GPT model + text generation.
I install dependencies and run training or generation with:
pip install -r requirements.txt
python train.py
python generate.py
I built the project on August 14, 2026.