@prefix this: . @prefix sub: . @prefix rdfs: . @prefix xsd: . @prefix np: . @prefix npx: . @prefix dcterms: . @prefix prov: . @prefix schema: . @prefix foaf: . @prefix orcid: . sub:head { this: np:hasAssertion sub:assertion; np:hasProvenance sub:provenance; np:hasPublicationInfo sub:pubinfo; a np:Nanopublication . } sub:assertion { "preprint" . sub:assertion dcterms:creator ; ; ; rdfs:comment """ To reduce evaluation contamination @XuanmingZhang07 @Zhou_Yu_AI @columbianlp et al. convert dataset examples into templates(Fig.) https://arxiv.org/abs/2406.17681 EWOK datasets are built to have this trait https://x.com/neuranna/status/1791465842632454184 Interesting trend will it last? solve contamination? https://twitter.com/LChoshen/status/1806396147281637645/photo/1 @XuanmingZhang07 @Zhou_Yu_AI @columbianlp If you ask me, a nice step, but it only solves the worst contamination (clear training on the test set). Not on just training on similar formats, synthetic data etc. to improve. So it is a good approach that should last, but we need more. (@deliprao you had similar claim right?) """; schema:keywords "dataset-templates", "evaluation-contamination", "ewok-datasets", "language-model-benchmarking", "varbench"; . "forumPost" . } sub:provenance { a prov:SoftwareAgent; prov:actedOnBehalfOf . sub:activity a ; prov:wasAssociatedWith . sub:assertion prov:linksTo ; prov:wasAssociatedWith ; prov:wasAttributedTo orcid:0000-0002-0085-6496, ; prov:wasGeneratedBy sub:activity . foaf:account orcid:0000-0002-0085-6496, . } sub:pubinfo { sub:sig npx:hasAlgorithm "RSA"; npx:hasPublicKey "MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEArHtI92jm8pAYVsvJabxLGfOT+7G0JyJGh2gwjB5x2pFPga6wWTd+rNBWWUZViIFnaJrBEsJpgdnoupLU9ppwn+khMiGRfxqGsDDzwHcj3Jc75CRys7d3etwXdBdoXfBgjsJiZBazwm13idr6tljRrC1TaEJBnRQAqzBw9cLDeGY77cSznzXT39feUGT168dpCSE9O6u/48DvvWVqciHGsH9cQ+LroJJVsMrorwtsdZnAK+q48wtIP6pIpw5shSJ5LnA0qeN/f4TvTFDV6ItYIXjiWWpTECc/Bxmfnyat3B5xWCu9nvz8fEs7Ns0TuzQwT3/K55iSKDEIi/E0nO97xwIDAQAB"; npx:hasSignature "IhJE7e2QCuhi9lBnA6zyNAFEnuD0Kq+6UXzbq4THcwqG0odW9IFvLzUeFrsO55KnIGA1Mz4O5TDx9CZvCLnkRxNYmCM4ikItw54oCAYwDE40zONWvcYAZpeSQvECknmIwaTEikPBENjFKF6BxgYEgtxCP0pMc37iuAvUHQq5uBBugkbSr8FgRFi3+3IIOEiiWANOxiioxtznNCG7VdaSnD1XkvGQGwoS7HxUdQCOvj+1cZBs4YnLk2ZAixo0AI0X9N3/ucT0om5uGRBtkMdluhggmrK7v6o4dxGDH6pV/RE5RpnxIjOcdtzczPF0MhpaLCZiPPYbzFkIb3WITivdmw=="; npx:hasSignatureTarget this:; npx:singedBy ; prov:wasAssociatedWith . this: dcterms:created "2024-09-12T18:53:31.101Z"^^xsd:dateTime; dcterms:creator ; dcterms:license ; npx:hasNanopubType ; npx:wasCreatedAt ; rdfs:label "CoSMO Semantic Post"; prov:wasAttributedTo orcid:0000-0002-0085-6496; "0xf6ECcfD463afB464dcC85b051DF2E93E2646E6D2" . foaf:account orcid:0000-0002-0085-6496; foaf:name "Leshem Choshen 🤖🤗 @ICML wanna talk?" . }