{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ACK6XXY55YGKRFFDPSCAWFALKA","short_pith_number":"pith:ACK6XXY5","schema_version":"1.0","canonical_sha256":"0095ebdf1dee0ca894a37c840b140b503691e9404008cd65a3fe5ab615f28d0c","source":{"kind":"arxiv","id":"2607.27248","version":1},"attestation_state":"computed","paper":{"title":"Divergence Decoding: Training-Free Capability Fusion","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Dechen Zhang, Fanyang Mo, Hao Li, He Cao, Li Yuan, Shuo Yang, Yimi Wang, Zhiyuan Yan, Ziang Wu","submitted_at":"2026-07-28T06:52:19Z","abstract_excerpt":"While large language models excel in reasoning, these generalists often lack knowledge for specialized scientific domains. Conversely, domain models~(specialists), while knowledgeable, suffer from specialization side-effects including diminished logic and reduced robustness.To address this dilemma, we introduce Divergence Decoding, a training-free framework for capability fusion. It reconstructs the \"draft-and-verify\" skeleton of speculative decoding into an adaptive routing mechanism. The core is using Jensen-Shannon divergence to monitor the distributional disagreement between the two models"},"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":"2607.27248","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-28T06:52:19Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5a9a001b64da32a7ca6d2beb4c3206adc42aa082dc040fb90678374b89892839","abstract_canon_sha256":"dfa0bf7e74dbe2bad808d0c17bc5403cd0babf46522ae4b6b7ed719bd3f79880"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0095ebdf1dee0ca894a37c840b140b503691e9404008cd65a3fe5ab615f28d0c","last_reissued_at":"2026-07-31T00:10:24.835767Z","signature_status":"unsigned_v0","first_computed_at":"2026-07-31T00:10:24.835767Z"},"graph_snapshot":{"paper":{"title":"Divergence Decoding: Training-Free Capability Fusion","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Dechen Zhang, Fanyang Mo, Hao Li, He Cao, Li Yuan, Shuo Yang, Yimi Wang, Zhiyuan Yan, Ziang Wu","submitted_at":"2026-07-28T06:52:19Z","abstract_excerpt":"While large language models excel in reasoning, these generalists often lack knowledge for specialized scientific domains. Conversely, domain models~(specialists), while knowledgeable, suffer from specialization side-effects including diminished logic and reduced robustness.To address this dilemma, we introduce Divergence Decoding, a training-free framework for capability fusion. It reconstructs the \"draft-and-verify\" skeleton of speculative decoding into an adaptive routing mechanism. The core is using Jensen-Shannon divergence to monitor the distributional disagreement between the two models"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.27248","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/2607.27248/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":"2607.27248","created_at":"2026-07-31T00:10:24.838049+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.27248v1","created_at":"2026-07-31T00:10:24.838049+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.27248","created_at":"2026-07-31T00:10:24.838049+00:00"},{"alias_kind":"pith_short_12","alias_value":"ACK6XXY55YGK","created_at":"2026-07-31T00:10:24.838049+00:00"},{"alias_kind":"pith_short_16","alias_value":"ACK6XXY55YGKRFFD","created_at":"2026-07-31T00:10:24.838049+00:00"},{"alias_kind":"pith_short_8","alias_value":"ACK6XXY5","created_at":"2026-07-31T00:10:24.838049+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/ACK6XXY55YGKRFFDPSCAWFALKA","json":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA.json","graph_json":"https://pith.science/api/pith-number/ACK6XXY55YGKRFFDPSCAWFALKA/graph.json","events_json":"https://pith.science/api/pith-number/ACK6XXY55YGKRFFDPSCAWFALKA/events.json","paper":"https://pith.science/paper/ACK6XXY5"},"agent_actions":{"view_html":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA","download_json":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA.json","view_paper":"https://pith.science/paper/ACK6XXY5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.27248&json=true","fetch_graph":"https://pith.science/api/pith-number/ACK6XXY55YGKRFFDPSCAWFALKA/graph.json","fetch_events":"https://pith.science/api/pith-number/ACK6XXY55YGKRFFDPSCAWFALKA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA/action/storage_attestation","attest_author":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA/action/author_attestation","sign_citation":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA/action/citation_signature","submit_replication":"https://pith.science/pith/ACK6XXY55YGKRFFDPSCAWFALKA/action/replication_record"}},"created_at":"2026-07-31T00:10:24.838049+00:00","updated_at":"2026-07-31T00:10:24.838049+00:00"}