GHGbench: A Unified Multi-Entity, Multi-Task Benchmark for Carbon Emission Prediction
Researchers have introduced GHGbench, a new open dataset and benchmark designed to improve the prediction of carbon emissions at both company and building levels. The benchmark includes extensive data for company emissions, incorporating financial and sectoral signals, alongside harmonized building-level data from multiple sources with climate and remote-sensing information. Initial findings highlight that predicting building emissions is more challenging than company emissions, and out-of-distribution performance significantly outweighs in-distribution gains, with multimodal embeddings proving useful where tabular generalization falters. AI
IMPACT Provides a standardized evaluation framework for AI models tackling climate change prediction.