EMLOPS Assignment 1

Tasks

• Change the train.py file to work with PyTorch
• Repeat the whole tutorial (Links to an external site.) again, but this time committing to GitHub.
• your data MUST not be there on GitHub, but only available to you locally
• when submitting, mention the accuracy of each model (trained twice, each time 10 epochs (or more)).
• You must publish metrics.csv file

Test Cases

test.py file checks

• You have NOT uploaded data.zip (check where it is expected)
• You have NOT uploaded model.h5 file (check where it is expected)
• Your accuracy of the model is more than 70% (read from metrics)
• The accuracy of your model for cat and dog is independently more than 70% (read from metrics.csv file) Share the GitHub link

Model

Augumentations

This was too cute to ignore

• Model 1 - From scratch

  • Metrics

Epochs	Test Accuracy	Dogs Accuracy	Cats Accuracy
9	62.4360	58.25	75.0
10	67.6242	36.75	83.0

• Commit Details - d2317d6 - Basic model, trained with 1000 images 59% acc failing test case
• As expected, this model was failing test cases, hence we need a new model

• Model 2 - Pretrained VGG16

  • Metrics

Epochs	Test Accuracy	Dogs Accuracy	Cats Accuracy
9	98.983	99.375	98.5185
10	98.6441	98.75	98.5185

• Latest Commit
• Model is passing all the test cases