phdhatamodel / README.md
msmaje's picture
Add comprehensive model card
f0a0ecb verified
metadata
language:
  - en
  - yo
  - ha
  - ig
  - sw
  - am
  - pcm
license: apache-2.0
base_model: davlan/afro-xlmr-base
tags:
  - text-classification
  - human-ai-text-attribution
  - hata
  - african-languages
  - multilingual
datasets:
  - msmaje/phd-hata-african-dataset
metrics:
  - accuracy
  - f1

AfroXLMR for Human-AI Text Attribution (HATA)

This model is a fine-tuned version of davlan/afro-xlmr-base for Human-AI Text Attribution in African languages.

Model Description

  • Model Type: Text Classification (Binary)
  • Base Model: AfroXLMR-base
  • Languages: Yoruba, Hausa, Igbo, Swahili, Amharic, Nigerian Pidgin, English
  • Task: Distinguishing between human-written and AI-generated text

Performance

Metric Score
Accuracy 1.0000
F1 Score 1.0000
Precision 1.0000
Recall 1.0000

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_name = "msmaje/phdhatamodel"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

text = "Your text here"
inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=128)

with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)
    predicted_class = torch.argmax(predictions, dim=-1).item()
    
labels = {0: "Human-written", 1: "AI-generated"}
print(f"Prediction: {labels[predicted_class]}")

Training Details

  • Dataset: msmaje/phd-hata-african-dataset
  • Training samples: 128,000
  • Validation samples: 32,000
  • Epochs: 3
  • Learning Rate: 2e-5
  • Batch Size: 16

Citation

@misc{msmaje2025hata,
  author = {Maje, M.S.},
  title = {AfroXLMR for Human-AI Text Attribution},
  year = {2025},
  publisher = {HuggingFace},
  url = {https://huggingface.co/msmaje/phdhatamodel}
}