🚧 Project in Progress 🚧 This repository is a personal learning journey into understanding how neural networks work. The approach is hands-on experimentation, where I implement and test different models and ideas.
- Build an intuitive understanding of neural networks.
- Compare coding styles and solutions from different AI assistants.
- Run experiments on real-world tasks where neural networks can be applied.
A comparison between neural network implementations written with the help of 3 different AI agents:
- Qwen3
- Claude Sonnet 4
- ChatGPT
The goal is to analyze differences in:
- Code style
- Clarity
- Performance
- Learning value
Applying neural networks to practical problems. The first experiment:
-
5-Star Review Classifier
- Input: review text + metadata
- Output: exact star rating (1–5)
Future experiments will be added step by step.
- Python
- PyTorch / TensorFlow (depending on experiment)
- Scikit-learn
- Jupyter notebooks for exploration
This project is exploratory. Expect code to be iterative, experimental, and sometimes messy. The focus is learning, not production-ready ML pipelines.
- Set up repo
- Complete Part 1 – Agent comparison
- Build baseline 5-star classifier
- Explore improvements with embeddings / advanced architectures
- Add more real-world experiments
This is primarily a personal learning project, but feedback and ideas are welcome! If you have suggestions for interesting experiments, feel free to open an issue.