{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:AKH3ZOAG563ILHTCKHK62KIULZ","short_pith_number":"pith:AKH3ZOAG","canonical_record":{"source":{"id":"2505.19327","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-25T21:26:18Z","cross_cats_sorted":[],"title_canon_sha256":"f047e66f996131566140068bdf5a488969f7e9932adc213fa767b553d04e7fd9","abstract_canon_sha256":"86eedfd008905fd00659c545bc80e8399b02f9980f7a4f714aea2ae64d29dc4e"},"schema_version":"1.0"},"canonical_sha256":"028fbcb806efb6859e6251d5ed29145e52ff2d9be3155d769430a016aa881459","source":{"kind":"arxiv","id":"2505.19327","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.19327","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"arxiv_version","alias_value":"2505.19327v1","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19327","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"pith_short_12","alias_value":"AKH3ZOAG563I","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"pith_short_16","alias_value":"AKH3ZOAG563ILHTC","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"pith_short_8","alias_value":"AKH3ZOAG","created_at":"2026-07-05T11:09:28Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:AKH3ZOAG563ILHTCKHK62KIULZ","target":"record","payload":{"canonical_record":{"source":{"id":"2505.19327","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-25T21:26:18Z","cross_cats_sorted":[],"title_canon_sha256":"f047e66f996131566140068bdf5a488969f7e9932adc213fa767b553d04e7fd9","abstract_canon_sha256":"86eedfd008905fd00659c545bc80e8399b02f9980f7a4f714aea2ae64d29dc4e"},"schema_version":"1.0"},"canonical_sha256":"028fbcb806efb6859e6251d5ed29145e52ff2d9be3155d769430a016aa881459","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:28.446293Z","signature_b64":"VBGWIadQ1aVdWr2T7t5KCdE8kU9WdajKxpmJrQp2TW8fs83klG7bvu5toDtY++ISKhYqyx6EkDwuGSdcUeB1Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"028fbcb806efb6859e6251d5ed29145e52ff2d9be3155d769430a016aa881459","last_reissued_at":"2026-07-05T11:09:28.445809Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:28.445809Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2505.19327","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:09:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"57uc1siw2z2OTE339HtO8L4RYPU2QmVEjo+tqU+yCaCvgpa/hJwVGhatRcXkq+L2RnbnNqPo2lM3aghsxaFXAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T19:55:16.370335Z"},"content_sha256":"0cd7d6473a48a50bf8bef08d74508d92bc0ab7948aa242266669505f173cfe22","schema_version":"1.0","event_id":"sha256:0cd7d6473a48a50bf8bef08d74508d92bc0ab7948aa242266669505f173cfe22"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:AKH3ZOAG563ILHTCKHK62KIULZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Paying Alignment Tax with Contrastive Learning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Antonio del Rio Chanona, Buse Sibel Korkmaz, Elizabeth M. Daly, Rahul Nair","submitted_at":"2025-05-25T21:26:18Z","abstract_excerpt":"Current debiasing approaches often result a degradation in model capabilities such as factual accuracy and knowledge retention. Through systematic evaluation across multiple benchmarks, we demonstrate that existing debiasing methods face fundamental trade-offs, particularly in smaller models, leading to reduced truthfulness, knowledge loss, or unintelligible outputs. To address these limitations, we propose a contrastive learning framework that learns through carefully constructed positive and negative examples. Our approach introduces contrast computation and dynamic loss scaling to balance b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19327","kind":"arxiv","version":1},"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/2505.19327/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":"2026-07-05T11:09:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"153ErhBsZXVFD0dhx1Aig4Dj+/oIF3v6f2Etqma47fd8o/ya8uRFr9W5EgDXjfzb0zofjrBPaLd95uoShzNuCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T19:55:16.370836Z"},"content_sha256":"a8935da73d6667faf0f0d820eb71b7ab15aa3a557b08274ae1f15903f7c15116","schema_version":"1.0","event_id":"sha256:a8935da73d6667faf0f0d820eb71b7ab15aa3a557b08274ae1f15903f7c15116"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/AKH3ZOAG563ILHTCKHK62KIULZ/bundle.json","state_url":"https://pith.science/pith/AKH3ZOAG563ILHTCKHK62KIULZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/AKH3ZOAG563ILHTCKHK62KIULZ/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":"2026-08-09T19:55:16Z","links":{"resolver":"https://pith.science/pith/AKH3ZOAG563ILHTCKHK62KIULZ","bundle":"https://pith.science/pith/AKH3ZOAG563ILHTCKHK62KIULZ/bundle.json","state":"https://pith.science/pith/AKH3ZOAG563ILHTCKHK62KIULZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/AKH3ZOAG563ILHTCKHK62KIULZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:AKH3ZOAG563ILHTCKHK62KIULZ","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":"86eedfd008905fd00659c545bc80e8399b02f9980f7a4f714aea2ae64d29dc4e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-25T21:26:18Z","title_canon_sha256":"f047e66f996131566140068bdf5a488969f7e9932adc213fa767b553d04e7fd9"},"schema_version":"1.0","source":{"id":"2505.19327","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2505.19327","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"arxiv_version","alias_value":"2505.19327v1","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19327","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"pith_short_12","alias_value":"AKH3ZOAG563I","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"pith_short_16","alias_value":"AKH3ZOAG563ILHTC","created_at":"2026-07-05T11:09:28Z"},{"alias_kind":"pith_short_8","alias_value":"AKH3ZOAG","created_at":"2026-07-05T11:09:28Z"}],"graph_snapshots":[{"event_id":"sha256:a8935da73d6667faf0f0d820eb71b7ab15aa3a557b08274ae1f15903f7c15116","target":"graph","created_at":"2026-07-05T11:09:28Z","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/2505.19327/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Current debiasing approaches often result a degradation in model capabilities such as factual accuracy and knowledge retention. Through systematic evaluation across multiple benchmarks, we demonstrate that existing debiasing methods face fundamental trade-offs, particularly in smaller models, leading to reduced truthfulness, knowledge loss, or unintelligible outputs. To address these limitations, we propose a contrastive learning framework that learns through carefully constructed positive and negative examples. Our approach introduces contrast computation and dynamic loss scaling to balance b","authors_text":"Antonio del Rio Chanona, Buse Sibel Korkmaz, Elizabeth M. Daly, Rahul Nair","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-25T21:26:18Z","title":"Paying Alignment Tax with Contrastive Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19327","kind":"arxiv","version":1},"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":"sha256:0cd7d6473a48a50bf8bef08d74508d92bc0ab7948aa242266669505f173cfe22","target":"record","created_at":"2026-07-05T11:09:28Z","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":"86eedfd008905fd00659c545bc80e8399b02f9980f7a4f714aea2ae64d29dc4e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-25T21:26:18Z","title_canon_sha256":"f047e66f996131566140068bdf5a488969f7e9932adc213fa767b553d04e7fd9"},"schema_version":"1.0","source":{"id":"2505.19327","kind":"arxiv","version":1}},"canonical_sha256":"028fbcb806efb6859e6251d5ed29145e52ff2d9be3155d769430a016aa881459","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"028fbcb806efb6859e6251d5ed29145e52ff2d9be3155d769430a016aa881459","first_computed_at":"2026-07-05T11:09:28.445809Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:09:28.445809Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"VBGWIadQ1aVdWr2T7t5KCdE8kU9WdajKxpmJrQp2TW8fs83klG7bvu5toDtY++ISKhYqyx6EkDwuGSdcUeB1Dg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:09:28.446293Z","signed_message":"canonical_sha256_bytes"},"source_id":"2505.19327","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0cd7d6473a48a50bf8bef08d74508d92bc0ab7948aa242266669505f173cfe22","sha256:a8935da73d6667faf0f0d820eb71b7ab15aa3a557b08274ae1f15903f7c15116"],"state_sha256":"eaf2f3fbeef94cbdfbf9c8c7d39d192a7f887c2b38ce63fa4ad599443d21b59e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"dF9fFk85frHgd04zyHn9XpE+eeQ52L2TQlqiszqHkS6fmxfJ9HkH2mRqJZEyIz3H+/zly5uyJ+KmUsJ89AvqCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T19:55:16.374377Z","bundle_sha256":"938304ac149fe2198e54fc67e17165b3f87d8c84546c1d97bb346aea3abd4350"}}