{"id":6568,"date":"2025-04-16T05:13:00","date_gmt":"2025-04-15T20:13:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=6568"},"modified":"2025-04-16T05:13:00","modified_gmt":"2025-04-15T20:13:00","slug":"olmotrace-tracing-language-model-outputs-back-to-trillions-of-training-tokens","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=6568","title":{"rendered":"OLMoTrace: Tracing Language Model Outputs Back to Trillions of Training Tokens\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>OLMoTrace: Tracing Language Model Outputs Back to Trillions of Training Tokens\u00a0<\/strong>[119.6]<br>OLMoTrace\u306f\u3001\u8a00\u8a9e\u30e2\u30c7\u30eb\u306e\u30a2\u30a6\u30c8\u30d7\u30c3\u30c8\u3092\u3001\u5b8c\u5168\u306b\u30de\u30eb\u30c1\u30c8\u30ea\u30eb\u306e\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u306b\u30ea\u30a2\u30eb\u30bf\u30a4\u30e0\u3067\u30c8\u30ec\u30fc\u30b9\u3059\u308b\u3002 OLMoTrace\u306f\u3001\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c6\u30ad\u30b9\u30c8\u30b3\u30fc\u30d1\u30b9\u5185\u306e\u8a00\u8a9e\u30e2\u30c7\u30eb\u51fa\u529b\u306e\u30bb\u30b0\u30e1\u30f3\u30c8\u3068\u30c9\u30ad\u30e5\u30e1\u30f3\u30c8\u306e\u5197\u9577\u306a\u4e00\u81f4\u3092\u898b\u3064\u3051\u3001\u8868\u793a\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2504.07096v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2504.07096v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Wed, 09 Apr 2025 17:59:35 GMT)<\/li>\n\n\n\n<li>\u300cOLMOTRACE finds and shows verbatim matches between segments of language model output and documents in the training text corpora.\u300d\u3068\u3044\u3046\u30b7\u30b9\u30c6\u30e0\u306e\u63d0\u6848\u3068OSS\u5b9f\u88c5\u306e\u516c\u958b\u3002Limitation\u306b\u3082\u300cThe retrieved documents should not be interpreted as having a causal effect on the LM output, or as supporting evidence or citations for the LM output.\u300d\u3068\u66f8\u304b\u308c\u3066\u3044\u308b\u3068\u306f\u3044\u3048\uff08\u304b\u3064LLM\u306e\u30c7\u30fc\u30bf\u304c\u5fc5\u8981\u3068\u306f\u3044\u3048\uff09\u3001\u69d8\u3005\u306a\u5fdc\u7528\u304c\u8003\u3048\u3089\u308c\u305d\u3046\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/allenai\/infinigram-api\">GitHub &#8211; allenai\/infinigram-api<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[452],"class_list":["post-6568","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-xai"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6568","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6568"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/6568\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6568"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6568"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6568"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}