{"paper":{"title":"Hugging Carbon: Quantifying the Training Carbon Emissions of AI Models at Scale","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A FLOPs-based framework with tiered metadata handling estimates that training popular open-source models on Hugging Face has emitted approximately 58,000 metric tons of carbon and introduces the ATCI metric for training efficiency.","cross_cats":[],"primary_cat":"cs.CY","authors_text":"Jing Qiu, Jinjin Gu, Junhua Zhao, Ruibo Ming, Xinlei Wang","submitted_at":"2026-05-02T17:32:56Z","abstract_excerpt":"The scaling-law era has transformed artificial intelligence (AI) from research into a global industry, but its rapid growth also raises concerns over energy usage, carbon emissions, and environmental sustainability. Unlike traditional sectors, the AI industry still lacks systematic carbon accounting methods that support large-scale estimates without reproducing the original training process. This leaves open questions about how large the problem is today and how large it might be in the near future. Given its central role in hosting open-source AI models, the Hugging Face (HF) platform provide"},"claims":{"count":3,"items":[{"kind":"strongest_claim","text":"Our results show that training the most popular open-source models (with over 5,000 downloads) has resulted in approximately 5.8×10^4 metric tons of carbon emissions.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Given that the Hugging Face (HF) platform well represents the broader open-source community, we treat it as a large-scale, publicly accessible, and audit-ready corpus for carbon accounting.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A FLOPs-based framework with tiered metadata handling estimates that training popular open-source models on Hugging Face has emitted approximately 58,000 metric tons of carbon and introduces the ATCI metric for training efficiency.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"}],"snapshot_sha256":"6553e9ac33eb7289f9af845241b2a7fdb30c38aad1d040edaa6a12c2322be8b8"},"source":{"id":"2605.01549","kind":"arxiv","version":2},"verdict":{"id":"9dea25e1-dabf-4927-9ac9-387a5e8361a0","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-09T13:42:23.194793Z","strongest_claim":"Our results show that training the most popular open-source models (with over 5,000 downloads) has resulted in approximately 5.8×10^4 metric tons of carbon emissions.","one_line_summary":"A FLOPs-based framework with tiered metadata handling estimates that training popular open-source models on Hugging Face has emitted approximately 58,000 metric tons of carbon and introduces the ATCI metric for training efficiency.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Given that the Hugging Face (HF) platform well represents the broader open-source community, we treat it as a large-scale, publicly accessible, and audit-ready corpus for carbon accounting.","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.01549/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T17:40:03.132799Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"f4615f7312722b4f1e756d38c1fee5d5cd04d711a77bc9872ab8d4491bb7201f"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}