Hananguyen12/hana-cv-qa-enhanced

Enhanced Hana Nguyen CV Question-Answering Model

This is an enhanced version of the CV QA model with improved accuracy and performance.

Model Performance

• Average Confidence Score: 0.3131
• High Confidence Answers (>0.7): 20.0%
• Base Model: RoBERTa-base fine-tuned on SQuAD 2.0
• Training Data: 150+ augmented question-answer pairs

Key Improvements

1. Enhanced Context Structure: Better organized CV information
2. Data Augmentation: Question paraphrasing and variations
3. Better Answer Positioning: Improved answer location in context
4. Advanced Training: Custom optimizer, learning rate scheduling
5. Comprehensive Evaluation: Multiple performance metrics

Usage

from transformers import pipeline

qa_pipeline = pipeline("question-answering", model="Hananguyen12/hana-cv-qa-enhanced")

result = qa_pipeline(
    question="What is Hana's current research focus?",
    context=context
)

Training Details

• Epochs: 5 with early stopping
• Batch Size: 4 (with gradient accumulation)
• Learning Rate: 3e-5 with warmup
• Optimization: AdamW with differential learning rates
• Mixed Precision: FP16 training enabled

Supported Question Types

• Personal information queries
• Technical skills and proficiency levels
• Work experience and achievements
• Educational background
• Research and projects
• Contact information