Tudorx95/resnet18-cifar10-pytorch

ResNet18 pentru CIFAR-10

Model ResNet18 adaptat si antrenat pe CIFAR-10.

Performanta

Metrica	Valoare
Accuracy	0.8807
Precision	0.8823
Recall	0.8807
F1-Score	0.8806

Utilizare cu template_antrenare_pytorch.py

# In template, seteaza:
HUGGINGFACE_REPO_ID = "Tudorx95/resnet18-cifar10-pytorch"
MODEL_FILENAME = "ResNet18_CIFAR10.pth"

# Modelul se incarca automat in create_model()

Utilizare directa

import torch
import torchvision
import torch.nn as nn

# Incarca checkpoint-ul
checkpoint = torch.load('ResNet18_CIFAR10.pth', map_location='cpu')

# Reconstruieste arhitectura din config
arch = checkpoint['architecture']
model = torchvision.models.resnet18(weights=None)
mod = arch['modifications']
model.conv1 = nn.Conv2d(3, 64, **mod['conv1'])
model.maxpool = nn.Identity()
model.fc = nn.Linear(mod['fc']['in_features'], mod['fc']['out_features'])

# Incarca ponderile
model.load_state_dict(checkpoint['model_state_dict'])
model.eval()

Clase CIFAR-10

0: airplane, 1: automobile, 2: bird, 3: cat, 4: deer, 5: dog, 6: frog, 7: horse, 8: ship, 9: truck

Antrenare

• Epochs: 10
• Batch Size: 128
• Learning Rate: 0.001
• Optimizer: Adam
• Scheduler: StepLR (step=5, gamma=0.5)
• Augmentare: RandomCrop, RandomHorizontalFlip