The open-source website classification model `website_classification` can accurately identify website categories with an accuracy rate of 95.04%.

Website Classification

Developed by alimazhar-110

A website classification model based on DistilBERT, achieving an accuracy of 95.04% on an unknown dataset through fine-tuning.

Text Classification

Transformers

Open Source License:Apache-2.0 #High-precision text classification #Lightweight BERT #Multi-category website recognition

Downloads 3,844

Release Time : 2/2/2023

Model Overview

This is a website classification model based on natural language processing. By fine-tuning distilbert-base-uncased, it can efficiently and accurately classify websites.

Model Features

Efficient and accurate

Achieved an accuracy of 95.04% and an F1 value of 94.89 on the evaluation set, showing excellent performance.

Lightweight

Based on the DistilBERT architecture, it is lighter than the original BERT model and has a faster inference speed.

Optimized training

Using the Adam optimizer and a linear learning rate scheduler, the best performance was achieved after 30 rounds of training.

Model Capabilities

Website text classification

Natural language understanding

Text feature extraction

Use Cases

Content management

Automatic website classification

Automatically classify website content into predefined categories

Accuracy 95.04%

Network security

Malicious website recognition

Identify potential malicious websites by analyzing website content

🚀 website_classification

This model is a fine - tuned version of DistilBERT, designed to classify websites with high accuracy.

🚀 Quick Start

This model is a fine - tuned version of [distilbert - base - uncased](https://huggingface.co/distilbert - base - uncased) on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.2292
Accuracy: 0.9504
F1: 0.9489
Precision: 0.9510
Recall: 0.9504

📚 Documentation

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e - 05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon = 1e - 08
lr_scheduler_type: linear
num_epochs: 30
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
2.404	1.0	71	1.7840	0.8865	0.8776	0.8785	0.8865
1.295	2.0	142	0.8539	0.8972	0.8871	0.8803	0.8972
0.6186	3.0	213	0.4818	0.9326	0.9263	0.9266	0.9326
0.3103	4.0	284	0.3101	0.9397	0.9343	0.9324	0.9397
0.1618	5.0	355	0.3001	0.9291	0.9251	0.9278	0.9291
0.0893	6.0	426	0.2743	0.9291	0.9251	0.9276	0.9291
0.0547	7.0	497	0.2605	0.9255	0.9236	0.9334	0.9255
0.028	8.0	568	0.2167	0.9397	0.9375	0.9403	0.9397
0.0186	9.0	639	0.2096	0.9468	0.9467	0.9499	0.9468
0.0134	10.0	710	0.2219	0.9362	0.9354	0.9402	0.9362
0.0107	11.0	781	0.2124	0.9468	0.9466	0.9507	0.9468
0.0087	12.0	852	0.2119	0.9504	0.9497	0.9534	0.9504
0.0075	13.0	923	0.2141	0.9504	0.9497	0.9534	0.9504
0.0066	14.0	994	0.2198	0.9433	0.9415	0.9442	0.9433
0.0058	15.0	1065	0.2188	0.9468	0.9454	0.9474	0.9468
0.0052	16.0	1136	0.2181	0.9468	0.9454	0.9474	0.9468
0.0047	17.0	1207	0.2220	0.9504	0.9489	0.9510	0.9504
0.0044	18.0	1278	0.2232	0.9504	0.9489	0.9510	0.9504
0.004	19.0	1349	0.2216	0.9539	0.9535	0.9565	0.9539
0.0037	20.0	1420	0.2251	0.9504	0.9489	0.9510	0.9504
0.0036	21.0	1491	0.2275	0.9468	0.9451	0.9470	0.9468
0.0034	22.0	1562	0.2264	0.9539	0.9535	0.9565	0.9539
0.0032	23.0	1633	0.2283	0.9504	0.9489	0.9510	0.9504
0.003	24.0	1704	0.2299	0.9504	0.9489	0.9510	0.9504
0.0029	25.0	1775	0.2282	0.9468	0.9451	0.9470	0.9468
0.0029	26.0	1846	0.2288	0.9468	0.9451	0.9470	0.9468
0.0028	27.0	1917	0.2286	0.9504	0.9489	0.9510	0.9504
0.0027	28.0	1988	0.2293	0.9504	0.9489	0.9510	0.9504
0.0026	29.0	2059	0.2291	0.9504	0.9489	0.9510	0.9504
0.0026	30.0	2130	0.2292	0.9504	0.9489	0.9510	0.9504

Framework versions

Transformers 4.26.0
Pytorch 1.13.1+cu116
Datasets 2.9.0
Tokenizers 0.13.2

📄 License

This model is licensed under the Apache - 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご