This model is a fine-tuned version of Llama 3B, optimized for generating high-quality multiple-choice questions (MCQs) from input text. It combines the powerful language understanding capabilities of Llama with specialized training for educational content generation.

Model Details

Model Description

This model is designed to automatically generate multiple-choice questions from input text, making it particularly useful for educators, content creators, and educational technology platforms.

Developed by: Ahmed Othman
Model type: Fine-tuned Language Model
Language(s): English
License: Apache 2.0
Finetuned from model: meta-llama/Llama-3.2-3B-Instruct

Uses

Direct Use

The model can be used directly for:

Generating multiple-choice questions from educational texts
Creating assessment materials
Automated quiz generation
Educational content development

python

from transformers import AutoModelForSeq2SeqLM, AutoTokenizer

model = AutoModelForSeq2SeqLM.from_pretrained("AhmedOthman/Llama-3B-QA-Enhanced")
tokenizer = AutoTokenizer.from_pretrained("AhmedOthman/Llama-3B-QA-Enhanced")

text = "Your input text here"
inputs = tokenizer(text, return_tensors="pt", max_length=512, truncation=True)
outputs = model.generate(inputs.input_ids)
mcq = tokenizer.decode(outputs[0], skip_special_tokens=True)

Out-of-Scope Use

This model should not be used for:

Generating factually incorrect or misleading questions
Creating questions about sensitive or controversial topics
Replacing human expertise in high-stakes assessment development

Training Details

Training Data

The model was trained on a combination of:

SQuAD (Stanford Question Answering Dataset)
RACE (ReAding Comprehension from Examinations)

Training Procedure

Training Hyperparameters

Training regime: fp16 mixed precision
Maximum sequence length: 512 tokens
Learning rate: 2e-5
Batch size: 16
Number of epochs: 3

Evaluation

Metrics

The model was evaluated using:

BLEU score for question generation quality
ROUGE score for answer relevance
Accuracy of generated distractors
Human evaluation for question quality

Limitations and Bias

Limited to English language content
May generate simpler questions for complex topics
Performance varies with input text quality
May reflect biases present in training data

Environmental Impact

Base Model: Llama 3B
Fine-tuning Hardware: Single A100 GPU
Training Time: Approximately 8 hours

Citation

If you use this model in your research, please cite:

bibtex

@misc{othman2024llama3bqa,
  author = {Othman, Ahmed},
  title = {Llama-3B-QA-Enhanced},
  year = {2024},
  publisher = {HuggingFace},
  howpublished = {\url{https://huggingface.co/AhmedOthman/Llama-3B-QA-Enhanced}}
}

Model Card Contact

For questions or issues, please contact Ahmed Othman through the HuggingFace model repository.


This model card provides comprehensive information about your fine-tuned model while maintaining a professional and informative tone. You can further customize it by:

1. Adding specific performance metrics from your evaluation
2. Including more example outputs
3. Detailing any specific preprocessing steps used
4. Adding links to related research or projects

Would you like me to modify any particular section or add more specific details?