Graf
PulseAugur coverage of Graf — every cluster mentioning Graf across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
Attention ResUNet model updated for improved hourly precipitation forecasts
Researchers have updated their Attention Residual U-Net model, which refines precipitation forecasts. The updated model now uses a single trained model per season instead of 192 per-month, per-lead checkpoints, extendin…
-
New algorithmic framework uses LLMs to optimize crowdsourcing contest selection
Researchers have developed GRAF, a new algorithmic framework designed to optimize worker self-selection in crowdsourcing contests. GRAF aims to ensure that important contests receive adequate participation and effort, w…
-
New AI method improves hourly precipitation forecasts
Researchers have developed a new method called "Attention Residual U-Net" for probabilistic quantitative precipitation forecasting. This neural network is trained on numerical weather prediction data from The Weather Co…