---
language:
- en
license: apache-2.0
tags:
- text2text-generation
- paraphraser
- ai-humanizer
- t5
pipeline_tag: text2text-generation
---

# Advanced AI Humanizer & Neural Paraphraser Models Suite

This repository contains production-ready Neural Seq2Seq Transformer models designed for the **AI Humanizer Web Application**.

## Features:
- **Academic Writing Paraphrasing**: Restructures complex discourse into published academic style.
- **AI Detector Neutralizer**: Dismantles formulaic AI patterns, robotic phrasing, and unnatural sentence uniformities.
- **Fact & Citation Preservation**: Fully compatible with regex FactShields and math tokenization.
- **High-Speed Inference**: Supports PyTorch and ONNX Runtime execution.

## Usage with Hugging Face Transformers:
```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("Haiderali1212/ai-humanizer-models")
model = AutoModelForSeq2SeqLM.from_pretrained("Haiderali1212/ai-humanizer-models")

text = "paraphrase: Artificial intelligence plays a crucial role in modern technological developments. </s>"
inputs = tokenizer(text, padding="longest", return_tensors="pt")
outputs = model.generate(**inputs, num_beams=4, max_length=128)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```