Expired50% confidenceFactExact time
The metrics output by E2AM include energy consumption (Wh), carbon emissions (gCO₂eq region-aware), GPU utilization, FLOPs/MACs, latency, energy consumption per sample, accuracy per joule, and EAG.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/12/2026
First Seen
Valid until: 10/10/2026(expired)
Sources
E2AM: An Open-Source Tool to Quantify AI Training Energy Consumption and Carbon Emissions in Two Lines of Code
redditr/deeplearning7/11/2026
Related Claims
Unverified能量模型（EBM）的思想是为输入空间中的每个配置分配一个标量能量值，低能量对应高兼容性的输入组合63% similarUnverified同一模型在不同数据中心训练碳排放量可能相差40倍以上，冰岛碳强度约18 gCO₂eq/kWh，依赖煤电地区可能超过800 gCO₂eq/kWh55% similarUnverifiedA study based on a global energy–economy integrated model concluded that productivity gains brought about by AI may drive a net increase in global CO₂ emissions.55% similarUnverified代谢功率（Metabolic Power）指标由意大利学者Osgnach等人于2010年提出，通过加速度变化推算能量消耗55% similarUnverifiedThe vast majority of electrical energy consumed by data centers ultimately converts to heat and is released into the environment.54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/491599API
curl https://kongchang.com/api/v1/knowledge/claims/491599MCP
get_claim(id=491599)