{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:GYDOAY274I5E6SAP252JVT4W67","short_pith_number":"pith:GYDOAY27","schema_version":"1.0","canonical_sha256":"3606e0635fe23a4f480fd7749acf96f7cc83aaf6c8f98dde1cafa4c27b922f3b","source":{"kind":"arxiv","id":"2607.18304","version":1},"attestation_state":"computed","paper":{"title":"TD-DPO: Difference-Aware Preference Optimization for Mitigating Sycophancy in Clinical Autism Intervention Dialogue","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chenxi Li, Gang Pan, Haifeng Li, Junhong Lai, Lin Yao, Qing Zhou, Shuzhong Lai, Yueming Wang","submitted_at":"2026-07-17T05:25:48Z","abstract_excerpt":"The sycophancy of large language models can increase the safety risk in intervention dialogue for autistic children. Supervised fine-tuning can somewhat reduce sycophancy, but relying solely on positive examples is often insufficient to identify and correct failure patterns. We observe that sycophancy behaviors can often be localized to a limited span within the model response. In this regime, sequence-level preference optimization can over-update preference-irrelevant tokens and degrade intervention ability. To address this, we propose the \\textbf{M}inimal \\textbf{E}dit \\textbf{D}ata \\textbf{"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2607.18304","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-07-17T05:25:48Z","cross_cats_sorted":[],"title_canon_sha256":"bde5b365bbfd6e1939b1f097912dfff6b7ff1c1cb13d597e39ceb9747db3a023","abstract_canon_sha256":"ddf4b926fad45500bc8175a81b860a83f81f893e2258c93c9d61ab81181f24dd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-22T00:22:40.862615Z","signature_b64":"LOIQ3Ac0J40XqHITV6M/hQ5P4D2khEy2L66PUZCC7m1lvvOYwIBjA85YBoDkU76fxlBa2u98JQ7z5I4InMHqCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3606e0635fe23a4f480fd7749acf96f7cc83aaf6c8f98dde1cafa4c27b922f3b","last_reissued_at":"2026-07-22T00:22:40.861831Z","signature_status":"signed_v1","first_computed_at":"2026-07-22T00:22:40.861831Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TD-DPO: Difference-Aware Preference Optimization for Mitigating Sycophancy in Clinical Autism Intervention Dialogue","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chenxi Li, Gang Pan, Haifeng Li, Junhong Lai, Lin Yao, Qing Zhou, Shuzhong Lai, Yueming Wang","submitted_at":"2026-07-17T05:25:48Z","abstract_excerpt":"The sycophancy of large language models can increase the safety risk in intervention dialogue for autistic children. Supervised fine-tuning can somewhat reduce sycophancy, but relying solely on positive examples is often insufficient to identify and correct failure patterns. We observe that sycophancy behaviors can often be localized to a limited span within the model response. In this regime, sequence-level preference optimization can over-update preference-irrelevant tokens and degrade intervention ability. To address this, we propose the \\textbf{M}inimal \\textbf{E}dit \\textbf{D}ata \\textbf{"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.18304","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/2607.18304/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2607.18304","created_at":"2026-07-22T00:22:40.862255+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.18304v1","created_at":"2026-07-22T00:22:40.862255+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.18304","created_at":"2026-07-22T00:22:40.862255+00:00"},{"alias_kind":"pith_short_12","alias_value":"GYDOAY274I5E","created_at":"2026-07-22T00:22:40.862255+00:00"},{"alias_kind":"pith_short_16","alias_value":"GYDOAY274I5E6SAP","created_at":"2026-07-22T00:22:40.862255+00:00"},{"alias_kind":"pith_short_8","alias_value":"GYDOAY27","created_at":"2026-07-22T00:22:40.862255+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67","json":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67.json","graph_json":"https://pith.science/api/pith-number/GYDOAY274I5E6SAP252JVT4W67/graph.json","events_json":"https://pith.science/api/pith-number/GYDOAY274I5E6SAP252JVT4W67/events.json","paper":"https://pith.science/paper/GYDOAY27"},"agent_actions":{"view_html":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67","download_json":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67.json","view_paper":"https://pith.science/paper/GYDOAY27","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.18304&json=true","fetch_graph":"https://pith.science/api/pith-number/GYDOAY274I5E6SAP252JVT4W67/graph.json","fetch_events":"https://pith.science/api/pith-number/GYDOAY274I5E6SAP252JVT4W67/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67/action/storage_attestation","attest_author":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67/action/author_attestation","sign_citation":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67/action/citation_signature","submit_replication":"https://pith.science/pith/GYDOAY274I5E6SAP252JVT4W67/action/replication_record"}},"created_at":"2026-07-22T00:22:40.862255+00:00","updated_at":"2026-07-22T00:22:40.862255+00:00"}