vit-base-patch16-224-in21k-ucSat Open Source Image Classification Model

Vit Base Patch16 224 In21k Ucsat

Developed by YKXBCi

Image classification model based on Vision Transformer architecture, fine-tuned on an unknown dataset

Downloads 31

Release Time : 7/1/2022

Model Overview

This model is a fine-tuned version of the google/vit-base-patch16-224-in21k pre-trained model, primarily used for image classification tasks.

High accuracy

Achieves 96.88% accuracy and 99.31% Top-3 accuracy on the validation set

Efficient training

Requires only 4 training epochs to achieve high performance

Mixed precision training

Uses mixed_float16 precision for training, improving training efficiency

Image classification

Multi-category recognition

Top-3 prediction

Computer vision

General image classification

Classifies and recognizes input images

Validation accuracy 96.88%

Train Loss	Train Accuracy	Train Top - 3 - accuracy	Validation Loss	Validation Accuracy	Validation Top - 3 - accuracy	Epoch
2.7376	0.5375	0.7284	2.3789	0.8958	0.9757	0
2.1030	0.9449	0.9972	1.8664	0.9479	0.9896	1
1.6719	0.9812	1.0	1.5763	0.9618	0.9931	2
1.4357	0.9926	1.0	1.4201	0.9688	0.9931	3
1.3216	0.9960	1.0	1.3683	0.9688	0.9931	4

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base