{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:WEDH3WNY44OKNI6LV7JED7C3K7","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":"52e1a106602410e5a820cefad8871ee84ab7c69eb4bb4245082ca069a35b7624","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-09-29T17:53:09Z","title_canon_sha256":"5a75cc940a70669a15a011662be42fe98e8dd36c2d7302ac1b57040c2d83c0fa"},"schema_version":"1.0","source":{"id":"2509.25148","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.25148","created_at":"2026-06-19T16:12:15Z"},{"alias_kind":"arxiv_version","alias_value":"2509.25148v2","created_at":"2026-06-19T16:12:15Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.25148","created_at":"2026-06-19T16:12:15Z"},{"alias_kind":"pith_short_12","alias_value":"WEDH3WNY44OK","created_at":"2026-06-19T16:12:15Z"},{"alias_kind":"pith_short_16","alias_value":"WEDH3WNY44OKNI6L","created_at":"2026-06-19T16:12:15Z"},{"alias_kind":"pith_short_8","alias_value":"WEDH3WNY","created_at":"2026-06-19T16:12:15Z"}],"graph_snapshots":[{"event_id":"sha256:9e15ca71394903b1041d12da942c12fb5509d07e1b0fc0811592803adb9373a5","target":"graph","created_at":"2026-06-19T16:12:15Z","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/2509.25148/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Post-training alignment of large language models often combines supervised fine-tuning (SFT) on expert demonstrations with reinforcement learning (RL) from preference or verifiable feedback. SFT provides a useful behavioral anchor but can overfit to static demonstrations, whereas RL encourages exploration but may drift from expert behavior or exploit imperfect rewards. We propose \\textbf{AAPA} (\\emph{Adversarially Anchored Preference Alignment}), a plug-in framework that augments existing post-training objectives with a sentence-level adversarial anchoring signal. AAPA compares policy rollouts","authors_text":"Faqiang Qian, Kang An, Liangjian Wen, Mengya Gao, Weikun Zhang, Xuhui Zheng, Yichao Wu, Yong Dai, Ziliang Wang","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-09-29T17:53:09Z","title":"AAPA: Adversarially Anchored Preference Alignment for Post-Training of Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.25148","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:b69d0b4ce8c01b471150c1ca517ef67ccd353bfa3d0b7c1c04999ef80e73f6eb","target":"record","created_at":"2026-06-19T16:12:15Z","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":"52e1a106602410e5a820cefad8871ee84ab7c69eb4bb4245082ca069a35b7624","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-09-29T17:53:09Z","title_canon_sha256":"5a75cc940a70669a15a011662be42fe98e8dd36c2d7302ac1b57040c2d83c0fa"},"schema_version":"1.0","source":{"id":"2509.25148","kind":"arxiv","version":2}},"canonical_sha256":"b1067dd9b8e71ca6a3cbafd241fc5b57e502a7077eeef75a30be02204546052d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b1067dd9b8e71ca6a3cbafd241fc5b57e502a7077eeef75a30be02204546052d","first_computed_at":"2026-06-19T16:12:15.287284Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-19T16:12:15.287284Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"/XW+mHMMe9YTc15cyswRsqatawCg/18LW7s8ZooyoMx6BuJDbmZhBv4/J9KJBqXXasI2Kyvs5YH4yuIVqapdDw==","signature_status":"signed_v1","signed_at":"2026-06-19T16:12:15.287734Z","signed_message":"canonical_sha256_bytes"},"source_id":"2509.25148","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b69d0b4ce8c01b471150c1ca517ef67ccd353bfa3d0b7c1c04999ef80e73f6eb","sha256:9e15ca71394903b1041d12da942c12fb5509d07e1b0fc0811592803adb9373a5"],"state_sha256":"6c7bd6cd49218fcf0eeb3d5a9cd81452a025217abe0663c3483e1a6abf530778"}