Update model card
Browse files
README.md
ADDED
|
@@ -0,0 +1,91 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
tags:
|
| 3 |
+
- image-classification
|
| 4 |
+
- cifar100
|
| 5 |
+
- resnet50-pytorch
|
| 6 |
+
- pytorch
|
| 7 |
+
datasets:
|
| 8 |
+
- cifar100
|
| 9 |
+
metrics:
|
| 10 |
+
- accuracy
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# CIFAR-100 resnet50-pytorch
|
| 14 |
+
|
| 15 |
+
## Model Description
|
| 16 |
+
|
| 17 |
+
resnet50-pytorch trained on CIFAR-100 dataset with advanced augmentation techniques.
|
| 18 |
+
|
| 19 |
+
### Model Architecture
|
| 20 |
+
- **Architecture**: resnet50-pytorch
|
| 21 |
+
- **Dataset**: CIFAR-100
|
| 22 |
+
- **Classes**: 100
|
| 23 |
+
|
| 24 |
+
### Training Configuration
|
| 25 |
+
- **Batch Size**: 128
|
| 26 |
+
- **Optimizer**: sgd (momentum=0.9, weight_decay=1e-3)
|
| 27 |
+
- **Scheduler**: onecycle
|
| 28 |
+
- **Augmentation**: HorizontalFlip, ShiftScaleRotate, Cutout, ColorJitter
|
| 29 |
+
- **MixUp**: Alpha=0.2
|
| 30 |
+
- **Label Smoothing**: 0.1
|
| 31 |
+
- **Mixed Precision**: True
|
| 32 |
+
- **Gradient Clipping**: 1.0
|
| 33 |
+
|
| 34 |
+
### Performance
|
| 35 |
+
- **Best Test Accuracy**: 23.80%
|
| 36 |
+
- **Total Epochs Trained**: 2
|
| 37 |
+
- **Final Train Accuracy**: 16.38%
|
| 38 |
+
- **Final Test Accuracy**: 23.80%
|
| 39 |
+
|
| 40 |
+
### Training History
|
| 41 |
+
- **Best Epoch**: 2
|
| 42 |
+
- **Train Loss**: 2.5457 → 2.3992
|
| 43 |
+
- **Test Loss**: 3.7171 → 2.1703
|
| 44 |
+
|
| 45 |
+
### Usage
|
| 46 |
+
|
| 47 |
+
```python
|
| 48 |
+
import torch
|
| 49 |
+
from huggingface_hub import hf_hub_download
|
| 50 |
+
|
| 51 |
+
# Download model
|
| 52 |
+
checkpoint_path = hf_hub_download(
|
| 53 |
+
repo_id="pandurangpatil/imagenet10trial",
|
| 54 |
+
filename="best_model.pth"
|
| 55 |
+
)
|
| 56 |
+
|
| 57 |
+
# Load checkpoint
|
| 58 |
+
checkpoint = torch.load(checkpoint_path, map_location='cpu', weights_only=False)
|
| 59 |
+
|
| 60 |
+
# Load model (you'll need to have the model definition)
|
| 61 |
+
# from models import get_model
|
| 62 |
+
# model = get_model('resnet50-pytorch', num_classes=100)
|
| 63 |
+
# model.load_state_dict(checkpoint['model_state_dict'])
|
| 64 |
+
# model.eval()
|
| 65 |
+
```
|
| 66 |
+
|
| 67 |
+
### Training Details
|
| 68 |
+
- **Dataset**: CIFAR-100 (50,000 train, 10,000 test)
|
| 69 |
+
- **Classes**: 100
|
| 70 |
+
- **Image Size**: 32×32
|
| 71 |
+
- **Normalization**: mean=(0.5071, 0.4865, 0.4409), std=(0.2673, 0.2564, 0.2761)
|
| 72 |
+
|
| 73 |
+
### Files
|
| 74 |
+
- `best_model.pth` - Best performing model checkpoint
|
| 75 |
+
- `training_curves.png` - Training/test accuracy and loss curves
|
| 76 |
+
- `lr_finder_plot.png` - Learning rate finder results
|
| 77 |
+
- `metrics.json` - Complete training history
|
| 78 |
+
- `config.json` - Hyperparameter configuration
|
| 79 |
+
|
| 80 |
+
### License
|
| 81 |
+
MIT
|
| 82 |
+
|
| 83 |
+
### Citation
|
| 84 |
+
```bibtex
|
| 85 |
+
@misc{resnet50-pytorch-cifar100,
|
| 86 |
+
title = {CIFAR-100 resnet50-pytorch},
|
| 87 |
+
year = {2025},
|
| 88 |
+
publisher = {HuggingFace},
|
| 89 |
+
url = {https://huggingface.co/pandurangpatil/imagenet10trial}
|
| 90 |
+
}
|
| 91 |
+
```
|