jajupmochi/swiss-river-network-benchmark

feat(workbench): enable CPU torch + neural quick-train (LSTM/Transformer) on the Space

feat(workbench): enable CPU torch + neural quick-train (LSTM/Transformer) on the Space

feat(workbench): enable CPU torch + neural quick-train (LSTM/Transformer) on the Space

fix(workbench): add three.js import map so 3D river network renders on HF Space

fix: radar nowcasting axes populated from per-station rows (audit 2026-07-04)

fix: guard _resolve_data for shallow /app path (Space RUNTIME_ERROR)

feat: model upload/inference/analysis + CPU/GPU detect; model upload disabled on hosted Space

feat: model upload/inference/analysis + CPU/GPU detect; model upload disabled on hosted Space

feat: drought.ch-style forecast outlook, three.js 3D river network, temperature map; full-width UI; pytest suite

feat: drought.ch-style forecast outlook, three.js 3D river network, temperature map; full-width UI; pytest suite

feat: radar overlay computed from uploaded per-station metrics

fix: drop show_api (removed in gradio 6.19); hide footer via CSS

Workbench v3: code-generated radar; train/eval CPU sandbox; upload routing+validation; coverage & sinusoidal-PE views; English-only + show_api=False; About/resources

Upgrade: Switzerland station map + river graph, raw series & seasonality, forecasting/noise tabs, results table, multi-format upload

remove stray __pycache__

Add working Gradio demo (window-length sweep + per-station error) over paper CSVs; drop stray file

Create visualize_sinusoidal.py

initial commit