gradcam v2: gradcam_swin_b.png

gradcam v2: gradcam_retfound.png

gradcam v2: gradcam_resnet50.png

gradcam v2: gradcam_resnet101.png

gradcam v2: gradcam_inception_v3.png

gradcam v2: gradcam_dinov2_l.png

gradcam v2: gradcam_densenet121.png

gradcam v2: gradcam_clip_openai.png

gradcam v2: README

Remove stale kfold/gradcam sections pending v2 regeneration

Remove stale gradcam/ (pre-v2 weights, will be regenerated)

Remove stale kfold/ (pre-v2 weights, will be regenerated)

Remove non-research HF push helper

Fix 95% CI in results table; add k-fold table + compute disclosure

Add CITATION.cff

Add numeric confusion matrix CSVs

Add per-class precision/recall/F1/support CSV

Add dataset class distribution + sample grid

Add Quick Inference, machine-readable artefacts, and compute disclosure

Add academic polish: inference example, hparams, CITATION.cff, dataset figures, per-class CSVs

Document analysis/, gradcam/, and kfold/ artefacts in README

Add Grad-CAM saliency maps for 6 CNN-class models

Add v1 5-fold cross-validation results

Add post-hoc analysis: confusion matrices, ROC, reliability, per-class F1, accuracy ranking

Add 9-model fundus benchmark: weights + results + splits + code + README

initial commit