SilvaXNet 🚀 (formerly SilvaNet)

Welcome to SilvaXNet, the next evolution of SilvaNet! Now, with built-in GPU acceleration via CuPy, SilvaXNet provides a seamless deep learning experience for both CPU and GPU users. Whether you're an AI enthusiast, researcher, or educator, this library offers a lightweight, intuitive, and educational deep learning framework that runs efficiently on both CPU (NumPy) and GPU (CuPy).

---

🚀 What's New?

At this moment, we have SilvaNet (CPU version), with the intent to extend to SilvaXNet (GPU version).

⚡ SilvaNet (CPU version, NumPy-based)

• Retains pure NumPy implementation for maximum portability
• Ideal for environments without GPU support

🔥 SilvaXNet (Upcoming GPU-accelerated version with CuPy)

• Planned CuPy integration for GPU acceleration
• Seamless NumPy ↔ CuPy tensor operations
• Optimized matrix operations for speedup

✅ Unified API for CPU & GPU (Future Feature)

• Effortless switching between SilvaNet (NumPy) and SilvaXNet (CuPy)
• API remains consistent for both backends

🔬 Planned Improvements for Neural Network Support

• Convolutional Layers (CNNs) optimized for GPU
• Better gradient computation with autograd
• Enhanced support for ANNs, RNNs, LSTMs, GRUs, and more!

---

🌟 Key Features (Current SilvaNet Version)

• Autograd Support: Automatic differentiation for smooth backpropagation.
• Deep Learning Layers:
  • Fully Connected (Dense) Layers
  • Recurrent Layers: RNN, LSTM, GRU
  • Convolutional Layers (Currently CPU-based)
• Loss Functions: Cross-Entropy, MSE, and more.
• Optimized Computation: NumPy-based operations for efficiency.
• Model Management: Save and load trained models seamlessly.

---

🚀 Getting Started

Here's a quick example using SilvaNet (CPU-only):

import numpy as np  # For CPU (SilvaNet)
from nn.Layers import Sequential, Dense
from nn.losses import CrossEntropyLoss
from nn.optimizer import SGD
from Network import NeuralNetwork

# Sample data
X_train = np.random.rand(100, 10)
y_train = np.random.randint(0, 2, size=(100, 1))

# Define the neural network
model = Sequential()
model.add(Dense(n_inputs=10, n_units=64, activation='relu'))
model.add(Dense(n_inputs=64, n_units=32, activation='relu'))
model.add(Dense(n_inputs=32, n_units=2))

# Loss function and optimizer
loss_fn = CrossEntropyLoss()
optimizer = SGD(parameters=model.get_parameters(), alpha=0.01)

# Train
nn = NeuralNetwork(model, loss_fn, optimizer)
nn.fit(X_train, y_train, epochs=100, batch_size=16)

---

⚙️ Installation

To install SilvaNet, clone the repository:

git clone https://github.com/silvaxxx1/SilvaXNet
cd SilvaXNet

For SilvaNet (CPU):

pip install -r requirements_cpu.txt  # NumPy-based

GPU support (SilvaXNet) is currently in development.

---

📚 Documentation

Check out the full API reference, guides, and tutorials here: Documentation

---

🤝 Contributing

We welcome contributions from the community! If you want to improve SilvaNet or help develop SilvaXNet, check out our contributing guidelines. We’d love to hear your feedback!

---

📝 License

SilvaNet is licensed under the MIT License. See the LICENSE file for more details.

---