Train_Anything

Welcome to Train_Anything – your launchpad to get you cracking with deep neural networks! This repository is a hands-on playground for building, training, and exploring a variety of neural network architectures, from classic ANNs to advanced RNNs and CNNs. Whether you're a beginner or looking to deepen your understanding, maybe this repo can help.

---

🧠 Model Notebooks

1. Artificial Neural Networks (ANN)

• File: Train_ANN.ipynb, Train_ANN_Scratch.ipynb
• What it does: Implements fully connected networks for classification tasks. Learn about layers, activations, loss functions, and optimization.
• Dataset Example: MNIST (handwritten digits)

2. Convolutional Neural Networks (CNN)

• File: Train_CNN.ipynb
• What it does: Focuses on image data, using convolutional layers to extract spatial features. Includes data augmentation and a cancer detection use case.
• Dataset Example: Custom image datasets (see notebook for details)

3. Recurrent Neural Networks (RNN)

• File: Train_RNN.ipynb
• What it does: Handles sequential data, maintaining memory of previous inputs. Example: character-level modeling on Java code.
• Dataset Example: Java.txt (character sequences)

4. Text Models (Feedforward for NLP)

• File: Train_text.ipynb
• What it does: Applies feedforward neural nets to text, showing the limitations before attention/transformers. Includes tokenization, encoding, and prediction.
• Dataset Example: train.txt, Shakespeare.txt, etc.

---

📚 Datasets

The datasets/ folder contains a variety of files for different tasks:

• Categorical: download Mnist from kaggle
• Text: Shakespeare.txt, Java.txt, Book.txt, Ideas.txt, names.txt
• Finance: Apple_Stock.xls
• Other: Add your own datasets to experiment!

---

📖 Theory

• Neural Extraction.pdf: For a deep theoretical background, check out this PDF included in the repo.

---

🚀 Transformers

Transformers are covered in my other repository introducing LLMs and their structure

---

🏁 Getting Started

1. Clone the repo and install the required Python packages (see individual notebooks for dependencies).
2. Open any notebook and start experimenting!
3. Swap in your own datasets or tweak the architectures to see what you can build.

---

Happy learning and building!

---

For questions, suggestions, contributions or learning roadmaps, feel free to open an issue or pull request or join my discord channel.