MobileNetV3 Small 050 LAMB
| Property | Value |
|---|---|
| Parameter Count | 1.6M |
| Model Type | Image Classification |
| License | Apache-2.0 |
| Framework | PyTorch (timm) |
| Input Size | 224 x 224 |
| Paper | Searching for MobileNetV3 |
What is mobilenetv3_small_050.lamb_in1k?
This is a highly efficient variant of MobileNetV3 Small, specifically designed for resource-constrained environments. The model has been trained on ImageNet-1k using an advanced LAMB optimizer recipe similar to "ResNet Strikes Back" but with 50% longer training duration and EMA weight averaging.
Implementation Details
The model implements a sophisticated training approach combining LAMB optimization with exponential decay learning rate scheduling and warmup periods. It achieves efficient performance through its compact architecture while maintaining reasonable accuracy on image classification tasks.
- Uses LAMB optimizer with EMA weight averaging
- Implements step-based learning rate scheduling with warmup
- Optimized for 224x224 input images
- Features 1.6M parameters with 0.9M activations
Core Capabilities
- Image classification on ImageNet-1k dataset
- Feature extraction for transfer learning
- Efficient inference on resource-constrained devices
- Real-time image processing capabilities
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its efficient architecture combined with modern training techniques like LAMB optimization and EMA weight averaging, making it particularly suitable for mobile and edge devices while maintaining good performance.
Q: What are the recommended use cases?
The model is ideal for mobile applications, edge devices, and scenarios requiring real-time image classification with limited computational resources. It's particularly well-suited for embedded systems and IoT applications where model size and inference speed are critical.
