GMixer-24/224 RA3
| Property | Value |
|---|---|
| Parameter Count | 24.7M |
| Model Type | Image Classification |
| License | Apache-2.0 |
| Image Size | 224x224 |
| GMACs | 5.3 |
What is gmixer_24_224.ra3_in1k?
GMixer-24/224 is a custom variant of the MLP-Mixer architecture developed within the TIMM (PyTorch Image Models) framework. This model represents an innovative approach to image classification by incorporating SwiGLU activation functions, designed specifically for processing 224x224 pixel images with high efficiency.
Implementation Details
The model architecture features 24.7M parameters and operates at 5.3 GMACs, making it relatively lightweight while maintaining strong performance. It implements a unique mixing strategy using SwiGLU activations, differentiating it from traditional MLP-Mixer models.
- Optimized for 224x224 input resolution
- Features 14.5M activations
- Utilizes SwiGLU activation functions
- Trained on ImageNet-1k dataset
Core Capabilities
- High-performance image classification
- Feature extraction for downstream tasks
- Efficient processing of standard resolution images
- Supports both classification and embedding generation
Frequently Asked Questions
Q: What makes this model unique?
This model's uniqueness lies in its custom implementation of the MLP-Mixer architecture with SwiGLU activation functions, offering a balance between computational efficiency and performance. With 24.7M parameters, it's positioned as a mid-sized model suitable for various image classification tasks.
Q: What are the recommended use cases?
The model is particularly well-suited for image classification tasks requiring 224x224 resolution inputs. It can be used both for direct classification and as a feature extractor for transfer learning applications. The model's efficiency makes it suitable for production environments where computational resources need to be balanced with performance.
