{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:K6TJGRV6ZMMYR7OQ2F3IAYB6WV","short_pith_number":"pith:K6TJGRV6","canonical_record":{"source":{"id":"2407.06551","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T05:16:22Z","cross_cats_sorted":[],"title_canon_sha256":"3166afac418cb36461da48a3701db7831f12cf2e4a9a61e596899475cbf61d16","abstract_canon_sha256":"a006a0a7f3462ccb020cf9ed3b775e305f82c1bbd8e50f391865dde0b8000259"},"schema_version":"1.0"},"canonical_sha256":"57a69346becb1988fdd0d17680603eb555d2683537c5594599bf14a2f00a1c9e","source":{"kind":"arxiv","id":"2407.06551","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.06551","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"arxiv_version","alias_value":"2407.06551v2","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.06551","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"pith_short_12","alias_value":"K6TJGRV6ZMMY","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"pith_short_16","alias_value":"K6TJGRV6ZMMYR7OQ","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"pith_short_8","alias_value":"K6TJGRV6","created_at":"2026-07-05T09:16:50Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:K6TJGRV6ZMMYR7OQ2F3IAYB6WV","target":"record","payload":{"canonical_record":{"source":{"id":"2407.06551","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T05:16:22Z","cross_cats_sorted":[],"title_canon_sha256":"3166afac418cb36461da48a3701db7831f12cf2e4a9a61e596899475cbf61d16","abstract_canon_sha256":"a006a0a7f3462ccb020cf9ed3b775e305f82c1bbd8e50f391865dde0b8000259"},"schema_version":"1.0"},"canonical_sha256":"57a69346becb1988fdd0d17680603eb555d2683537c5594599bf14a2f00a1c9e","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:16:50.935229Z","signature_b64":"emF/P1iOv68ogBwnlIz2B8O32wNYtEMwqBnraY6hqng6T1BQlWhyutVd4Mvmb1l6jf+88GvuhGhtBZafdVaYDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"57a69346becb1988fdd0d17680603eb555d2683537c5594599bf14a2f00a1c9e","last_reissued_at":"2026-07-05T09:16:50.934710Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:16:50.934710Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2407.06551","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:16:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Kl3IT84l7BY0tXcG7yKzB1ZjrIxipDmhIa3sPJp75Y3d3fp7fNMREeFKu+4yqSH6cMRZQxwcE/QqiTMx8N5GDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T14:15:28.154268Z"},"content_sha256":"3daf4851f00afd9e541ca3acac0bc15384ba06dccfca92a91e3420368bd899ef","schema_version":"1.0","event_id":"sha256:3daf4851f00afd9e541ca3acac0bc15384ba06dccfca92a91e3420368bd899ef"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:K6TJGRV6ZMMYR7OQ2F3IAYB6WV","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"OffsetBias: Leveraging Debiased Data for Tuning Evaluators","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Daeyoung Kim, Junsoo Park, Meiying Ren, Sanghyuk Choi, Seungyeon Jwa","submitted_at":"2024-07-09T05:16:22Z","abstract_excerpt":"Employing Large Language Models (LLMs) to assess the quality of generated responses, such as prompting instruct-tuned models or fine-tuning judge models, has become a widely adopted evaluation method. It is also known that such evaluators are vulnerable to biases, such as favoring longer responses. While it is important to overcome this problem, the specifics of these biases remain under-explored. In this work, we qualitatively identify six types of biases inherent in various judge models. We propose EvalBiasBench as a meta-evaluation collection of hand-crafted test cases for each bias type. A"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.06551","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2407.06551/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:16:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"S2aDp1iN+9fUHaNtR27OAYWAr9k7zQA+Hk2wJGY0bqWxmXmhvpgMpERKzOhi3yh4lyKcyz3DQCLYPnhSr2jZDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T14:15:28.154814Z"},"content_sha256":"9327a1972325eae99b7b5ebafa4a3e4a7237ae73915344b72fb3ebda624a98a4","schema_version":"1.0","event_id":"sha256:9327a1972325eae99b7b5ebafa4a3e4a7237ae73915344b72fb3ebda624a98a4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/K6TJGRV6ZMMYR7OQ2F3IAYB6WV/bundle.json","state_url":"https://pith.science/pith/K6TJGRV6ZMMYR7OQ2F3IAYB6WV/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/K6TJGRV6ZMMYR7OQ2F3IAYB6WV/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-09T14:15:28Z","links":{"resolver":"https://pith.science/pith/K6TJGRV6ZMMYR7OQ2F3IAYB6WV","bundle":"https://pith.science/pith/K6TJGRV6ZMMYR7OQ2F3IAYB6WV/bundle.json","state":"https://pith.science/pith/K6TJGRV6ZMMYR7OQ2F3IAYB6WV/state.json","well_known_bundle":"https://pith.science/.well-known/pith/K6TJGRV6ZMMYR7OQ2F3IAYB6WV/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:K6TJGRV6ZMMYR7OQ2F3IAYB6WV","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"a006a0a7f3462ccb020cf9ed3b775e305f82c1bbd8e50f391865dde0b8000259","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T05:16:22Z","title_canon_sha256":"3166afac418cb36461da48a3701db7831f12cf2e4a9a61e596899475cbf61d16"},"schema_version":"1.0","source":{"id":"2407.06551","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2407.06551","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"arxiv_version","alias_value":"2407.06551v2","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.06551","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"pith_short_12","alias_value":"K6TJGRV6ZMMY","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"pith_short_16","alias_value":"K6TJGRV6ZMMYR7OQ","created_at":"2026-07-05T09:16:50Z"},{"alias_kind":"pith_short_8","alias_value":"K6TJGRV6","created_at":"2026-07-05T09:16:50Z"}],"graph_snapshots":[{"event_id":"sha256:9327a1972325eae99b7b5ebafa4a3e4a7237ae73915344b72fb3ebda624a98a4","target":"graph","created_at":"2026-07-05T09:16:50Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2407.06551/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Employing Large Language Models (LLMs) to assess the quality of generated responses, such as prompting instruct-tuned models or fine-tuning judge models, has become a widely adopted evaluation method. It is also known that such evaluators are vulnerable to biases, such as favoring longer responses. While it is important to overcome this problem, the specifics of these biases remain under-explored. In this work, we qualitatively identify six types of biases inherent in various judge models. We propose EvalBiasBench as a meta-evaluation collection of hand-crafted test cases for each bias type. A","authors_text":"Daeyoung Kim, Junsoo Park, Meiying Ren, Sanghyuk Choi, Seungyeon Jwa","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T05:16:22Z","title":"OffsetBias: Leveraging Debiased Data for Tuning Evaluators"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.06551","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:3daf4851f00afd9e541ca3acac0bc15384ba06dccfca92a91e3420368bd899ef","target":"record","created_at":"2026-07-05T09:16:50Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"a006a0a7f3462ccb020cf9ed3b775e305f82c1bbd8e50f391865dde0b8000259","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-09T05:16:22Z","title_canon_sha256":"3166afac418cb36461da48a3701db7831f12cf2e4a9a61e596899475cbf61d16"},"schema_version":"1.0","source":{"id":"2407.06551","kind":"arxiv","version":2}},"canonical_sha256":"57a69346becb1988fdd0d17680603eb555d2683537c5594599bf14a2f00a1c9e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"57a69346becb1988fdd0d17680603eb555d2683537c5594599bf14a2f00a1c9e","first_computed_at":"2026-07-05T09:16:50.934710Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:16:50.934710Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"emF/P1iOv68ogBwnlIz2B8O32wNYtEMwqBnraY6hqng6T1BQlWhyutVd4Mvmb1l6jf+88GvuhGhtBZafdVaYDg==","signature_status":"signed_v1","signed_at":"2026-07-05T09:16:50.935229Z","signed_message":"canonical_sha256_bytes"},"source_id":"2407.06551","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3daf4851f00afd9e541ca3acac0bc15384ba06dccfca92a91e3420368bd899ef","sha256:9327a1972325eae99b7b5ebafa4a3e4a7237ae73915344b72fb3ebda624a98a4"],"state_sha256":"5b08b084942da08fa4e01eca76fac5dfafbd3a8384dfb045712c5e299abd2d83"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8vaePIJ9FvXGEflYE6UQb8k/3Z6TRP/b7G5K6zl8PHXiIl+57OqoIJIzi7I7j2tRryV6IzNi3FoC+JsJm1UiBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T14:15:28.159707Z","bundle_sha256":"1668b5214156883a99b45f3c9aa0fa538a7bdb359efece66f31f1266d4a8afa7"}}