Dingdong-Inc/FreshRetailNet-LT-Leaderboard

Paper-reported baselines, browser-only forecasting pre-evaluation, external submissions and manually reviewed community results.

See STATICGUIDECN.md. No backend, Docker, token or private repository is required. Public evaluation labels are used for local checks. Prediction files are not sent by this page. Configure an external submission channel in config.json.

Data attribution: derived evaluation references come from Dingdong-Inc/FreshRetailNet-LT, licensed CC BY 4.0. This package selects evaluation columns, derives historical mean sales, and converts the reference to browser JSON. Exact source revision and SHA256 fingerprints are in data/reference.json. The leaderboard baseline values are attributed to the supplied FreshRetailNet-LT manuscript; they are not newly reproduced model runs.