{"paper":{"title":"GHGbench: A Unified Multi-Entity, Multi-Task Benchmark for Carbon Emission Prediction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"GHGbench shows building carbon emissions are structurally harder to predict than company emissions, with out-of-distribution gaps dominating model differences.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chao Xue, Flora Salim, Lihuan Li, Siyuan Zheng, Yifan Duan","submitted_at":"2026-05-13T16:20:49Z","abstract_excerpt":"Open datasets and benchmarks for entity-level carbon-emission prediction remain fragmented across access, scale, granularity, and evaluation. We introduce GHGbench, an open dataset and benchmark for company- and building-level greenhouse-gas prediction. The company track contains 32,000+ company-year records from 12,000+ firms with Scope 1+2 and Scope 3 disclosures and financial/sectoral signals; the building track harmonises 491,591 building-year records from 13 open sources into a single schema across 26 metropolitan areas (10 U.S., 15 Australian, 1 Singaporean), with climate covariates and "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Three benchmark-level findings emerge: (i) building emissions are structurally harder than company emissions; (ii) the in-distribution to out-of-distribution gap dwarfs any within-model gap across both the company track and the building track, and a tabular foundation model is, to our knowledge, the first baseline to open a paired-bootstrap-significant gap over tuned trees on a multi-city building-emissions task; (iii) multimodal remote-sensing embeddings help precisely where tabular generalisation breaks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The harmonization of 13 heterogeneous building data sources into a single schema produces accurate, unbiased labels and features without introducing systematic errors that affect the reported generalization gaps.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"GHGbench is a new multi-entity benchmark for company- and building-level carbon emission prediction that shows building tasks are harder, out-of-distribution gaps dominate, and multimodal data aids generalization.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"GHGbench shows building carbon emissions are structurally harder to predict than company emissions, with out-of-distribution gaps dominating model differences.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"192f17b4bf4439ef7f2895da8736bf03552d8be1b3bdcf49e6523125b7febb9c"},"source":{"id":"2605.13743","kind":"arxiv","version":1},"verdict":{"id":"edbe21f9-71a3-4279-a244-1bd1c1a08e4e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:45:12.384933Z","strongest_claim":"Three benchmark-level findings emerge: (i) building emissions are structurally harder than company emissions; (ii) the in-distribution to out-of-distribution gap dwarfs any within-model gap across both the company track and the building track, and a tabular foundation model is, to our knowledge, the first baseline to open a paired-bootstrap-significant gap over tuned trees on a multi-city building-emissions task; (iii) multimodal remote-sensing embeddings help precisely where tabular generalisation breaks.","one_line_summary":"GHGbench is a new multi-entity benchmark for company- and building-level carbon emission prediction that shows building tasks are harder, out-of-distribution gaps dominate, and multimodal data aids generalization.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The harmonization of 13 heterogeneous building data sources into a single schema produces accurate, unbiased labels and features without introducing systematic errors that affect the reported generalization gaps.","pith_extraction_headline":"GHGbench shows building carbon emissions are structurally harder to predict than company emissions, with out-of-distribution gaps dominating model differences."},"references":{"count":46,"sample":[{"doi":"","year":2024,"title":"Maddix, Hao Wang, Michael W","work_id":"c3896363-cde8-417a-b93e-b8bfe3c2faeb","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"The Claude 3 model family: Opus, Sonnet, Haiku","work_id":"10cdd886-e474-49b9-9424-c263e80d32f4","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1016/j.apenergy.2020.115413","year":2020,"title":"EnergyStar++: Towards more accurate and explanatory building energy benchmarking.Applied Energy, 276:115413, 2020","work_id":"4234abcf-ba21-41c8-8234-5db7d72dcd2a","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.3390/su15043391","year":2023,"title":"Greenhouse gases emissions: Estimating corporate non-reported emissions using interpretable machine learning","work_id":"4204017e-c5cd-4e62-a73c-65843a3d4c6c","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1038/s41597-025-05664-8","year":2025,"title":"Addressing data gaps in sustainability reporting: A benchmark dataset for greenhouse gas emission extraction.Scientific Data, 12: 1497, 2025","work_id":"cda9ced3-8656-46bf-9704-704eeade1849","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":46,"snapshot_sha256":"b044601972f575a5f367813cf440bf4fd20019bfacd6444cbdbef5b6849ae661","internal_anchors":3},"formal_canon":{"evidence_count":2,"snapshot_sha256":"472aef6549dc78007ea98ee62019aec35ad67937efc0248299287e873cf81e7c"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}