Deep Learning Mini Project

Tasked with coming up with a modified residual network (ResNet) architecture with the highest test accuracy on the CIFAR-10 image classification dataset, under the constraint that your model has no more than 5 million parameters.

Results:

• Validation Accuracy - 94.59%
• Kaggle Custom dataset Accuracy - 84.2%

Wandb:

• Wandb first sweep
• Wandb second sweep

Contributors:

• Vishwa Gopalakrishnan
• Aditya Krishna
• Ohm Patel