DistilBERT AI-vs-Human Text Classifier
This is a binary text classification model built on top of distilbert-base-uncased.
It has been fine-tuned to distinguish between AI-generated and human-written text.
Base model: DistilBERT (uncased)
Task: Sequence classification
Labels:
0 → Human-written text
1 → AI-generated text
Training Details:
Model is fine-tuned on a small custom dataset of ~1.4k samples
Batch size: 16
Epochs: 10
Learning rate: 5e-6
Performance:
Best validation metrics:
Accuracy: 0.5730 (57.3%)
Precision: 0.6162
Recall: 0.9858
F1-score: 0.6814
Usage
Load the model and tokenizer with the Hugging Face Transformers library, provide a text input, and the model will output a label indicating whether the text is AI-generated or human-written.
Framework: PyTorch, Hugging Face Transformers
License: cc-by-nc-4.0
NOTE: This model is experimental and not intended for production use
