nerdy-commits/praxis-dr-grading

build: track gif files with Git LFS for Hugging Face Spaces compatibility

perf: cache graph stats and use approximate betweenness centrality in patient network tab

fix: CUDA fallback, install grad-cam+shap, add network stats panel to dashboard

Complete all portfolio gaps: add baseline comparison data/chart, integrate in Streamlit dashboard, export pruned PyVis patient network HTML, and resolve timm/pytest environment warnings

Pin numpy<2 for torch 2.1.0 compat; fix use_container_width deprecation

Fix numpy compat error, cache metadata/heatmaps, move cv2/PIL to top-level imports

Fix GradScaler import for torch 2.1.x and pre-warm model cache to stop shivering

Add libgl1 and libglib2.0-0 to Dockerfile to fix cv2 import error

Fix 403 upload error and flickering: disable CORS/XSRF, cache model loading, fix paths

Fix build: use requirements_hf.txt to avoid torch reinstall conflict

Optimize Dockerfile with CPU-only PyTorch to prevent build timeout

Remove software-properties-common from Dockerfile system deps

Fix Dockerfile to run as non-root USER user (UID 1000) for HF Spaces

Configure and prep Praxis for Hugging Face Spaces deployment with Docker and Git LFS

Initial commit after roadmap alignment and consistency fixes