Aitextdetector

🚀 AI Text Detector

This model is a fine-tuned version of GPT-2, designed for text classification.

🚀 Quick Start

This model supports three categories: AI, zhihu, and other, representing AI-generated text, Zhihu user answers, and other text respectively.

📦 Installation

No specific installation steps are provided in the original document, so this section is skipped.

💻 Usage Examples

Basic Usage

You can easily use the model through the pre - set text classification pipeline of transformers.pipeline:

from transformers import pipeline
classifier = pipeline("text-classification",model="./checkpoints",device="cuda")
text = "加密通信助长犯罪的说法并不成立，理由如下：隐私权的基本保障：加密技术保障了个人和组织的隐私权，这是基本人权的一部分。无论是个人的日常通信、商业机密，还是新闻记者的来源保护，加密通信都在确保合法和正当的隐私需求不被侵犯。"
print(classifier(text))

The output result is:

[{'label': 'AI', 'score': 0.9951270818710327}]

Advanced Usage

The method of using the model itself for inference:

import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("hugfaceguy0001/AITextDetector")
model = AutoModelForSequenceClassification.from_pretrained("hugfaceguy0001/AITextDetector").to("cuda")

def detect_text(text):
    with torch.no_grad():
        inputs = tokenizer(text, return_tensors="pt").to("cuda")
        logits = model(**inputs).logits
        predicted_class_id = logits.argmax().item()
    return model.config.id2label[predicted_class_id]

text = "黑神话这个游戏，你喷它优化、喷它地图、喷它空气墙、喷它战斗、喷它剧情；这些东西不管我自己觉得有没有喷一下的必要，可也都或多或少的能理解它为什么被喷的原因。"
print(detect_text(text))

The output result is:

zhihu

📚 Documentation

Training Data

Approximately 52,000 answer texts from alpaca-data-gpt4-chinese are used as AI-generated text, about 52,000 randomly selected answer texts from Zhihu-KOL are used as Zhihu user answer texts, and around 52,000 randomly selected texts from SkyPile-150B are used as other text.

A total of about 156,000 classified texts form the dataset, with 80% used for training and 20% for testing.

Performance on the Test Set

The classification accuracy is accuracy = 0.9802627363024672.

The confusion matrix composed of the true labels and detection results of each sample is as follows:

📄 License

The license of this model is OpenRAIL.

Additional Information