MultiLabel_V3 Open-Source Multi-Label Classification Model - Based on Fine-Tuning, Classification Accuracy Reaches 73.7%

Multilabel V3

Developed by Madronus

A multi-label classification model fine-tuned from google/vit-base-patch16-224-in21k with an accuracy of 73.7%

Downloads 19

Release Time : 3/14/2023

Model Overview

This model is a multi-label classification model based on the Vision Transformer architecture, suitable for image classification tasks.

High accuracy

Achieves 73.7% accuracy on the evaluation set

Based on ViT architecture

Utilizes the Vision Transformer base architecture with powerful feature extraction capabilities

Multi-label classification

Supports multi-label classification tasks for images

Image classification

Multi-label prediction

Image analysis

Multi-label image classification

Classify and label images containing multiple objects

Accuracy 73.7%

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8572	0.1	100	1.1607	0.6466
0.8578	0.2	200	1.1956	0.6499
0.7362	0.3	300	1.1235	0.6885
0.8569	0.39	400	1.0460	0.6891
0.4851	0.49	500	1.1213	0.6891
0.7252	0.59	600	1.1512	0.6720
0.6333	0.69	700	1.1039	0.6913
0.6239	0.79	800	1.0636	0.7001
0.2768	0.89	900	1.0386	0.7073
0.4872	0.99	1000	1.0311	0.7062
0.3049	1.09	1100	1.0437	0.7155
0.1435	1.18	1200	1.0343	0.7222
0.2088	1.28	1300	1.0784	0.7194
0.4972	1.38	1400	1.1072	0.7166
0.3604	1.48	1500	1.0438	0.7150
0.2726	1.58	1600	1.0077	0.7293
0.3106	1.68	1700	1.0029	0.7326
0.3259	1.78	1800	0.9906	0.7310
0.3323	1.88	1900	0.9729	0.7359
0.2998	1.97	2000	0.9683	0.7370

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base