|
|
--- |
|
|
library_name: keras |
|
|
--- |
|
|
|
|
|
## Model description |
|
|
|
|
|
This repo contains the trained model Self-supervised contrastive learning with SimSiam on Cifar 10 Dataset. |
|
|
Keras link: https://keras.io/examples/vision/simsiam/ |
|
|
|
|
|
## Intended uses & limitations |
|
|
The trained model can be used as a learned representation for downstream tasks like image classification. |
|
|
|
|
|
|
|
|
## Training and evaluation data |
|
|
|
|
|
Original Cifar 10 train & test dataset were loaded from tensorflow datasets. |
|
|
Two particular augmentation transforms that seem to matter the most are: |
|
|
1.) Random resized crops |
|
|
2.) Color distortions |
|
|
|
|
|
## Training procedure |
|
|
|
|
|
### Training hyperparameters |
|
|
|
|
|
The following hyperparameters were used during training: |
|
|
|
|
|
| name | learning_rate | decay | momentum | nesterov | training_precision | |
|
|
|----|-------------|-----|--------|--------|------------------| |
|
|
|SGD|{'class_name': 'CosineDecay', 'config': {'initial_learning_rate': 0.03, 'decay_steps': 3900, 'alpha': 0.0, 'name': None}}|0.0|0.8999999761581421|False|float32| |
|
|
|
|
|
## Model Plot |
|
|
|
|
|
<details> |
|
|
<summary>View Model Plot</summary> |
|
|
|
|
|
 |
|
|
|
|
|
</details> |
