Fix TypeError by reverting threshold parameter name back to threshold_ratio

Change attention crop to mask outside bounding boxes, fix overlapping text in zone reference

Increase font size for Zone Reference and Intensity Scale; increase upload limit to 10MB

Change BBox color to red and increase size; update GPT prompt for starting sentence and red box explanation

Fix attention bbox rendering: apply heatmap, increase box size/thickness. Remove text from UI.

Downgrade transformers to fix torch 2.4.1 compatibility on Hugging Face

Fix ModuleNotFoundError by adding matplotlib to requirements; add intensity scale bar to UI

Update explainability with attention bbox, crop, and zone reference; remove GradCAM from UI

chore: update favicon to use the specific branding image

fix(e2e): update history button selector and brand site with new favicon/meta

Enhance segmentation UI layout, implement local download, and improve diagnosis typography

Align segmentation mask post-processing with Kaggle implementation

Add engaging step-by-step processing loading screen for diagnosis and segmentation

Implement Dedicated Storage Client to resolve RLS bypass and history issues

Fix segmentation RLS path, implement clear history, and update UI styling

Fix RLS for segmentation upload, history loading, and add About page

HF configured

Fix: Remove deprecated resume_download argument from transformers

Fix: Make storage bucket public to prevent download errors

update api.ts for HF

update api.ts

Fix: Replace custom CORS with standard CORSMiddleware regex

Fix: Expand CORS allowed origins for Lovable

Fix: Upgrade torch and add missing sklearn/openai

Fix: Add scipy, skimage and downgrade numpy

Fix: requirements.txt encoding

Fix: Update apt-get package names

Fix: Add Hugging Face metadata

Deployment without binary files