Model Card for Model ID

This model is a small vgg16_bn trained on svhn.

• Test Accuracy: 0.961240012292563
• License: MIT

How to Get Started with the Model

Use the code below to get started with the model.

import detectors
import timm

model = timm.create_model("vgg16_bn_svhn", pretrained=True)

Training Data

Training data is svhn.

Training Hyperparameters

• config: scripts/train_configs/svhn.json

• model: vgg16_bn_svhn

• dataset: svhn

• batch_size: 128

• epochs: 300

• validation_frequency: 5

• seed: 1

• criterion: CrossEntropyLoss

• criterion_kwargs: {}

• optimizer: SGD

• lr: 0.01

• optimizer_kwargs: {'momentum': 0.9, 'weight_decay': 0.0005}

• scheduler: MultiStepLR

• scheduler_kwargs: {'gamma': 0.1, 'milestones': [75, 100, 150, 225]}

• debug: False

Testing Data

Testing data is svhn.

This model card was created by Eduardo Dadalto.