{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,19]],"date-time":"2026-03-19T22:55:00Z","timestamp":1773960900514,"version":"3.50.1"},"reference-count":23,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,11,25]],"date-time":"2025-11-25T00:00:00Z","timestamp":1764028800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,11,25]],"date-time":"2025-11-25T00:00:00Z","timestamp":1764028800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100011033","name":"Agencia Estatal de Investigaci\u00f3n","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100011033","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,11,25]]},"DOI":"10.1109\/fllm67465.2025.11391090","type":"proceedings-article","created":{"date-parts":[[2026,2,19]],"date-time":"2026-02-19T20:55:46Z","timestamp":1771534546000},"page":"961-967","source":"Crossref","is-referenced-by-count":0,"title":["The Generative Energy Arena (GEA): Incorporating Energy Awareness in Large Language Model Human Evaluations"],"prefix":"10.1109","author":[{"given":"Carlos","family":"Arriaga","sequence":"first","affiliation":[{"name":"Universidad Polit&#x00E9;cnica de Madrid,Information Processing and Telecommunications Center (IPTC),Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Gonzalo","family":"Mart\u00ednez","sequence":"additional","affiliation":[{"name":"Universidad Polit&#x00E9;cnica de Madrid,Information Processing and Telecommunications Center (IPTC),Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Eneko","family":"Send\u00edn","sequence":"additional","affiliation":[{"name":"Universidad Polit&#x00E9;cnica de Madrid,Information Processing and Telecommunications Center (IPTC),Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Javier","family":"Conde","sequence":"additional","affiliation":[{"name":"Universidad Polit&#x00E9;cnica de Madrid,Information Processing and Telecommunications Center (IPTC),Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pedro","family":"Reviriego","sequence":"additional","affiliation":[{"name":"Universidad Polit&#x00E9;cnica de Madrid,Information Processing and Telecommunications Center (IPTC),Spain"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1145\/3641289"},{"key":"ref2","article-title":"Measuring massive multitask language understanding","author":"Hendrycks","year":"2020"},{"key":"ref3","article-title":"Large language models are not robust multiple choice selectors","volume-title":"The Twelfth International Conference on Learning Representations","author":"Zheng"},{"key":"ref4","article-title":"Benchmark data contamination of large language models: A survey","author":"Xu","year":"2024"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.70777\/si.v2i1.13973"},{"key":"ref6","article-title":"Judging llm-as-a-judge with mt-bench and chatbot arena","volume":"36","author":"Zheng","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref7","article-title":"Llm evaluators recognize and favor their own generations","author":"Panickssery","year":"2024"},{"key":"ref8","article-title":"Chatbot arena: An open platform for evaluating llms by human preference","author":"Chiang","year":"2024"},{"key":"ref9","article-title":"Improving your model ranking on chatbot arena by vote rigging","author":"Min","year":"2025"},{"key":"ref10","article-title":"Holistically evaluating the environmental impact of creating language models","author":"Morrison","year":"2025"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.acl-long.519"},{"key":"ref12","article-title":"Investigating energy efficiency and performance trade-offs in llm inference across tasks and dvfs settings","author":"Maliakel","year":"2025"},{"key":"ref13","article-title":"The ML.ENERGY Benchmark: Toward automated inference energy measurement and optimization","author":"Chung","year":"2025"},{"key":"ref14","article-title":"Genaiarena: An open evaluation platform for generative models","author":"Jiang","year":"2024","journal-title":"NeurIPS 2024 Track on Datasets and Benchmarks"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.acl-long.223"},{"key":"ref16","article-title":"am-elo: A stable framework for arena-based llm evaluation","author":"Liu","year":"2025"},{"key":"ref17","article-title":"Projecting the electricity demand growth of generative ai large language models in the us","year":"2025"},{"key":"ref18","article-title":"How hungry is ai? benchmarking energy, water, and carbon footprint of llm inference","author":"Jegham","year":"2025"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/MC.2024.3399384"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/HPEC58863.2023.10363447"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1145\/3381831"},{"key":"ref22","article-title":"Inference economics of language models","author":"Erdil","year":"2025"},{"key":"ref23","article-title":"Cost-of-pass: An economic framework for evaluating language models","author":"Erol","year":"2025"}],"event":{"name":"2025 3rd International Conference on Foundation and Large Language Models (FLLM)","location":"Vienna, Austria","start":{"date-parts":[[2025,11,25]]},"end":{"date-parts":[[2025,11,28]]}},"container-title":["2025 3rd International Conference on Foundation and Large Language Models (FLLM)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11390736\/11390873\/11391090.pdf?arnumber=11391090","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,19]],"date-time":"2026-03-19T20:03:21Z","timestamp":1773950601000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11391090\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,11,25]]},"references-count":23,"URL":"https:\/\/doi.org\/10.1109\/fllm67465.2025.11391090","relation":{},"subject":[],"published":{"date-parts":[[2025,11,25]]}}}