vit-base-patch16-224-Futurama_Image_multilabel_clf

This model is a fine-tuned version of google/vit-base-patch16-224.

It achieves the following results on the evaluation set:

• Loss: 0.0592
• F1: 0.9818
• Roc Auc: 0.9842
• Accuracy: 0.9672

Model description

This is a multilabel classification model of screenshot images from the show Futurama.

For more information on how it was created, check out the following link: https://github.com/DunnBC22/Vision_Audio_and_Multimodal_Projects/blob/main/Computer%20Vision/Image%20Classification/Multilabel%20Classification/Futurama%20Screenshots/Futurama%20-%20ML%20Image%20CLF.ipynb

Intended uses & limitations

This model is intended to demonstrate my ability to solve a complex problem using technology.

Training and evaluation data

Dataset Source: https://www.kaggle.com/datasets/gonzalorecioc/futurama-frames-with-characteronscreen-data

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.2456	1.0	916	0.0723	0.9711	0.9746	0.9481
0.0269	2.0	1832	0.0545	0.9799	0.9818	0.9640
0.0086	3.0	2748	0.0580	0.9794	0.9814	0.9623
0.0044	4.0	3664	0.0612	0.9814	0.9832	0.9651
0.0027	5.0	4580	0.0592	0.9818	0.9842	0.9672
0.0017	6.0	5496	0.0634	0.9800	0.9832	0.9645
0.0012	7.0	6412	0.0657	0.9817	0.9840	0.9667
0.0005	8.0	7328	0.0668	0.9812	0.9836	0.9667

Framework versions

• Transformers 4.26.1
• Pytorch 1.12.1
• Datasets 2.8.0
• Tokenizers 0.12.1