Wei Yang (William) Submit HW7

[williamgoodhouses-rgb]

Week 7 Homework Submission – Synthetic QA & Fine-Tuning

This PR submits my Week 7 homework for MLE in the Generative AI Era, demonstrating an end-to-end pipeline for synthetic QA generation and supervised fine-tuning.

Included Notebooks

WeiYang-HW7-submit-v3.ipynb (local)

PDF ingestion, chunking, and synthetic QA generation using a local LLM

Dataset preparation and inspection

WeiYang_HW7_submit_colab_v2.ipynb (Colab T4 GPU)

Loads prepared synthetic QA data

Runs lightweight supervised fine-tuning using Hugging Face / TRL

Notes

Data generation (local) and training (Colab) are intentionally separated for environment stability

Training is kept minimal (1 epoch) to demonstrate correctness and reproducibility

A README is included with execution instructions and design rationale

Thank you for reviewing!