bharathwajmanoharan / aqp_generator Goto Github PK

• Utilizing the compact yet mighty T5 Small model for tackling PDF comprehension and question generation tasks.
• Harnessing its prowess to grasp technical jargon and intricate details.

• Employing the ADAMW optimizer to fine-tune model parameters, ensuring optimal performance.
• Tweaking the learning rate to enhance model adaptability to technical content.

• Delving into the world of question generation, we prompt the model to craft insightful queries from PDF text.
• Evaluating the quality of generated questions using industry-standard metrics.

• BLEU Score: Elevating comprehension to new heights, scoring a commendable 0.75.
• Rouge Score: Exhibiting solid performance with 0.65/0.80, hinting at room for further improvement.
• BERTScore: Striking gold with an impressive 0.85, setting the benchmark high.
• METEOR Score: Clocking in at 0.78, showcasing consistent proficiency.

• Charting a course for refining ROUGE score recall for a more comprehensive evaluation.
• Exploring ensemble techniques to amplify model capabilities and student engagement.

The fusion of T5 Small's agility with the precision of AdamW optimizer heralds a new era in PDF understanding and question crafting. As students dive into complex texts, this dynamic duo stands ready to illuminate pathways to knowledge with crisp comprehension and thought-provoking queries. With each evaluation metric setting a new standard, the journey towards enriched learning experiences continues, fueled by innovation and driven by excellence.

Together, let's continue to explore, learn, and grow, knowing that the best is yet to come!