 |
NeuZephyr
Simple DL Framework
|
|
NeuZephyr
Simple DL Framework
|
NeuZephyr is a lightweight deep learning library developed in C++ with CUDA C, designed to provide efficient GPU acceleration for deep learning model training and inference. Its goal is to help developers quickly implement deep learning models while maintaining an easy-to-use interface.
▌ Core Starting Point
👉 Begin with the foundational nz::Model class documentation! This class enables you to:
‣ 🧱 Define neural network architectures
‣ ⚙️ Manipulate model structures
‣ 🚂 Unify training/inference workflows
▌ Seamless Expansion Path
After mastering nz::Model, effortlessly integrate:
│
├ 🔥 Tensors - Data representation
├ 🧠 Nodes - Computational units
├ 🎯 Optimizers - Learning strategies
└ ...and other advanced features!
<INSTALL_DIR>/NeuZephyr/bin to your system's PATH environment variable.NeuZephyr in your CMake project: A Tensor is the fundamental data structure in NeuZephyr. It represents multi-dimensional arrays of data, which are essential for storing and manipulating data in deep learning models. Tensors are used in all computations, including matrix operations, activation functions, and more.
Tensors in NeuZephyr support a wide range of operations such as addition, multiplication, and broadcasting. The nz::data::Tensor class is designed to be efficient and is optimized for use with CUDA, ensuring that tensor operations are performed on the GPU for faster computation.
For more detailed documentation on how to use the Tensor class, see the Tensor.cuh file.
A Node in NeuZephyr represents a unit of computation in the computational graph. It can be an input/output (I/O) node, an operation node (such as matrix multiplication), or a loss function node (e.g., Mean Squared Error or Cross-Entropy). Nodes connect together to form a computational graph, which is used to define the flow of data and operations in a neural network.
Nodes are defined within the nz::nodes namespace. Each type of node has specific functionality, and they work together to perform the computations required for training and inference.
For more detailed documentation on the different node types, refer to the Nodes.cuh file or the nz::nodes namespace documentation.
An Optimizer is responsible for adjusting the parameters of the model (such as weights and biases) during training to minimize the loss function. NeuZephyr includes several commonly used optimizers, such as Stochastic Gradient Descent (SGD) and Adam.
Optimizers play a crucial role in the training process, and different optimizers can be more suitable for different tasks. The nz::opt namespace contains classes for implementing various optimization algorithms.
For detailed documentation on each optimizer and how to configure them, please refer to the Optimizer.cuh file or the nz::opt namespace documentation.
The ComputeGraph is the backbone of any deep learning model in NeuZephyr. It defines the structure of the neural network, including the layers, operations, and data flow. Each node in the graph represents a computation, and the connections between them define the execution order.
If you are looking to build your own neural network using NeuZephyr, you will need to understand how to create and manipulate a ComputeGraph. This class allows you to define complex models and perform training or inference on them. It also handles automatic differentiation for backpropagation during training.
For a deeper dive into how to use the ComputeGraph class and build custom neural networks, refer to the documentation for the ComputeGraph.cuh file.
For more detailed information, please refer to the corresponding header files or the documentation within each class/namespace. If you are new to NeuZephyr, we recommend starting with the Tensor and ComputeGraph documentation to get an understanding of how the core components work together.
For further assistance, feel free to check the repository or reach out to the maintainers.
For inquiries, issues, or contributions, please contact:
This project is provided strictly for learning and research purposes. It is not intended for production use or commercial deployment.
This project is licensed under the MIT License.
See the LICENSE file for details.
1.12.0