|
|
--- |
|
|
license: cc |
|
|
tags: |
|
|
- onnx |
|
|
- image-classification |
|
|
- cifar10 |
|
|
- dropout |
|
|
- aidge |
|
|
pipeline_tag: image-classification |
|
|
datasets: |
|
|
- cifar10 |
|
|
metrics: |
|
|
- type: accuracy |
|
|
value: 69.96% |
|
|
model-index: |
|
|
- name: Custom ResNet-18 with Integrated Dropout |
|
|
results: |
|
|
- task: |
|
|
type: image-classification |
|
|
name: Image Classification |
|
|
dataset: |
|
|
name: CIFAR-10 |
|
|
type: cifar10 |
|
|
metrics: |
|
|
- type: accuracy |
|
|
value: 69.96% |
|
|
--- |
|
|
|
|
|
# MarwaNet (ONNX) |
|
|
|
|
|
This is a **custom convolutional neural network (CNN)** trained on the **CIFAR-10** dataset, developed to test the integration of a **custom Dropout operator** for the **Aidge** platform. |
|
|
|
|
|
## Details |
|
|
|
|
|
- **Architecture**: Custom Convolutional Neural Network (CNN) with a Dropout layer |
|
|
- **Trained on**: CIFAR-10 (60,000 32x32 color images, 10 classes) |
|
|
- **Data Normalization**: `mean = [0.4914, 0.4822, 0.4465]` ; `std = [0.2023, 0.1994, 0.2010]` |
|
|
- **Dropout Probability**: 0.3 |
|
|
- **ONNX opset version**: 15 |
|
|
- **Conversion tool**: PyTorch → ONNX |
|
|
|
|
|
|
|
|
|
|
|
|
