memoraith 0.2.0

Advanced lightweight model profiler for deep learning frameworks

Memoraith

Memoraith is a cutting-edge, lightweight model profiler for deep learning frameworks, developed by Mehdi El Jouhfi. It's designed to revolutionize the optimization of neural network models by providing unparalleled insights into their performance characteristics.

Features

• Advanced support for PyTorch and TensorFlow models
• High-precision profiling of memory usage (CPU and GPU)
• Microsecond-accurate computation time measurement for each layer
• Sophisticated bottleneck and anomaly detection algorithms
• Generation of comprehensive, interactive reports with advanced visualizations
• Real-time visualization capabilities with minimal overhead
• Flexible programmatic and command-line interfaces

Installation

Install Memoraith using pip:

pip install memoraith

For GPU support and additional features:

pip install memoraith[full]

Quick Start

Here's an example of Memoraith in action with a PyTorch model:

from memoraith import profile_model, set_output_path
import torch
import torch.nn as nn

set_output_path('profiling_results/')

class AdvancedNet(nn.Module):
    def __init__(self):
        super(AdvancedNet, self).__init__()
        self.conv1 = nn.Conv2d(3, 64, kernel_size=3, padding=1)
        self.conv2 = nn.Conv2d(64, 128, kernel_size=3, padding=1)
        self.fc = nn.Linear(128 * 8 * 8, 10)

    def forward(self, x):
        x = torch.relu(self.conv1(x))
        x = torch.relu(self.conv2(x))
        x = x.view(x.size(0), -1)
        return self.fc(x)

@profile_model(memory=True, computation=True, gpu=True, network=True)
def train_model(model):
    optimizer = torch.optim.Adam(model.parameters())
    for _ in range(100):
        input_data = torch.randn(32, 3, 32, 32)
        output = model(input_data)
        loss = output.sum()
        loss.backward()
        optimizer.step()

if __name__ == "__main__":
    model = AdvancedNet()
    train_model(model)

This will generate a comprehensive profiling report in the 'profiling_results/' directory.

Documentation

For detailed information on Memoraith's advanced features, please refer to our comprehensive documentation.

Contributing

Contributions are welcome! Please read our Contributing Guide for details on our code of conduct and the process for submitting pull requests.

License

Memoraith is released under the MIT License. See the LICENSE file for more details.

Support

If you encounter any issues or have questions, please file an issue on the GitHub issue tracker.

Citing Memoraith

If you use Memoraith in your research, please cite it as follows:

@software{memoraith,
  author = {El Jouhfi, Mehdi},
  title = {Memoraith: Advanced Lightweight Model Profiler for Deep Learning},
  year = {2024},
  url = {https://github.com/mehdi342/Memoraith}
}

Contact

For inquiries, please contact Mehdi El Jouhfi at [e-mail].