Flan T5 Fine-Tuning on BioSummDataset S45893623

[jac0be]

FLAN-T5 + LoRA Layperson Radiology Summarisation

This PR adds my implementation for Project 13: fine-tuning FLAN-T5-base with LoRA to translate expert radiology reports into layperson-friendly summaries using the BioLaySumm 2025 dataset.

What's included

• modules.py — FLAN-T5 + LoRA model setup
• dataset.py — dataset loader with optional 90:10 train/test split
• train.py — full training loop, logging, ROUGE evaluation and model checkpointing
• predict.py — inference script for arbitrary reports
• eval.py — builds CSVs and plots from the training logs
• README.md — detailed description of the approach, results, metrics, and error analysis

Notes

• All code runs on a single GPU (tested on an RTX 5070 Ti).
• Final model achieves strong ROUGE-Lsum performance on the validation set.
• README includes dependency list, usage instructions, plots, and error analysis examples.

Please let me know if there's any changes you'd prefer. I'm more than happy to make them.

[gayanku]

Marking

Good/OK/Fair Practice (Design/Commenting, TF/Torch Usage)
	Good design and implementation.
	Spacing and comments.
	No Header blocks.	-1
Recognition Problem
	Good solution to problem.
	Driver Script present.
	File structure present.
	Good Usage & Demo & Visualisation & Data usage.
	Module present.
	Commenting present.
	No Data leakage found.
	Difficulty : Hard. Hard Difficulty : LLM
Commit Log
	Good Meaningful commit messages.
	Good Progressive commits.
Documentation
	Readme :Good.
	Model/technical explanation :Good.
	Description and Comments :Good.
	Markdown used. PDF NOT submitted.	-2
Pull Request
	Successful Pull Request (Working Algorithm Delivered on Time in Correct Branch).
	No Feedback required.
	Request Description is good.
TOTAL		-3

Marked as per the due date and changes after which aren't necessarily allowed to contribute to grade for fairness.
Subject to approval from Shakes