RegNetY-002 PyTorch Classification Model
| Property | Value |
|---|---|
| Parameter Count | 3.18M |
| License | MIT |
| Framework | PyTorch (timm) |
| Paper | Designing Network Design Spaces |
What is regnety_002.pycls_in1k?
RegNetY-002 is a lightweight convolutional neural network model designed for image classification tasks. It's part of the RegNet family, specifically optimized for efficiency with only 3.18M parameters and 0.2 GMACs computational requirement.
Implementation Details
This implementation is part of the timm library, featuring several enhancements over the original RegNet architecture:
- Stochastic depth for improved regularization
- Gradient checkpointing for memory efficiency
- Layer-wise learning rate decay
- Configurable output stride with dilation
- Flexible activation and normalization layers
Core Capabilities
- Image classification on 224x224 pixel inputs
- Feature extraction with multiple output scales
- Embedding generation for downstream tasks
- Efficient inference with low computational requirements
Frequently Asked Questions
Q: What makes this model unique?
RegNetY-002 stands out for its excellent efficiency-to-performance ratio, making it suitable for resource-constrained environments while maintaining reasonable accuracy levels. It's an ideal choice for mobile and edge devices.
Q: What are the recommended use cases?
This model is best suited for: 1) General image classification tasks requiring a lightweight model, 2) Feature extraction for transfer learning, and 3) Embedded systems with limited computational resources.
