{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:K2BTAWY7XEM4DSRLAYPH5T6WAP","short_pith_number":"pith:K2BTAWY7","schema_version":"1.0","canonical_sha256":"5683305b1fb919c1ca2b061e7ecfd603d4edfa2cc7d20a9e20a799c7efd3b8bb","source":{"kind":"arxiv","id":"2603.12996","version":2},"attestation_state":"computed","paper":{"title":"DAPD: Dependency-Aware Parallel Decoding via Attention for Diffusion LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Albert No, Bumjun Kim, Dongjae Jeon, Moongyu Jeon","submitted_at":"2026-03-13T13:52:02Z","abstract_excerpt":"Parallel decoding for Diffusion LLMs (dLLMs) is difficult because each denoising step provides only token-wise marginal distributions, while unmasking multiple tokens simultaneously requires accounting for inter-token dependencies. We propose Dependency-Aware Parallel Decoding (DAPD), a simple, training-free decoding method that uses self-attention to induce a conditional dependency graph over masked tokens. At each iteration, edges in this graph capture strong token interactions, while non-edges indicate weak dependence. Parallel decoding is then reduced to selecting an independent set on the"},"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":"2603.12996","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-03-13T13:52:02Z","cross_cats_sorted":[],"title_canon_sha256":"394e405f25926d00421f0ae043118dd82e8a8adb69c5c538864b10d730d50c74","abstract_canon_sha256":"dbf1743927708782bda36236e7fb5cdbf36010102d289a4a6fd3cc64627ca575"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T03:04:40.250480Z","signature_b64":"cREW4YWA2Yr/tdIIJ3NEGOx5g5sCAJYb+/4XKOaOlrcckiJt/KxklRPCP6cGVzHSQIAlWQOHKKGo4f/CXaeHDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5683305b1fb919c1ca2b061e7ecfd603d4edfa2cc7d20a9e20a799c7efd3b8bb","last_reissued_at":"2026-06-02T03:04:40.249959Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T03:04:40.249959Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DAPD: Dependency-Aware Parallel Decoding via Attention for Diffusion LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Albert No, Bumjun Kim, Dongjae Jeon, Moongyu Jeon","submitted_at":"2026-03-13T13:52:02Z","abstract_excerpt":"Parallel decoding for Diffusion LLMs (dLLMs) is difficult because each denoising step provides only token-wise marginal distributions, while unmasking multiple tokens simultaneously requires accounting for inter-token dependencies. We propose Dependency-Aware Parallel Decoding (DAPD), a simple, training-free decoding method that uses self-attention to induce a conditional dependency graph over masked tokens. At each iteration, edges in this graph capture strong token interactions, while non-edges indicate weak dependence. Parallel decoding is then reduced to selecting an independent set on the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.12996","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/2603.12996/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":"2603.12996","created_at":"2026-06-02T03:04:40.250022+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.12996v2","created_at":"2026-06-02T03:04:40.250022+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.12996","created_at":"2026-06-02T03:04:40.250022+00:00"},{"alias_kind":"pith_short_12","alias_value":"K2BTAWY7XEM4","created_at":"2026-06-02T03:04:40.250022+00:00"},{"alias_kind":"pith_short_16","alias_value":"K2BTAWY7XEM4DSRL","created_at":"2026-06-02T03:04:40.250022+00:00"},{"alias_kind":"pith_short_8","alias_value":"K2BTAWY7","created_at":"2026-06-02T03:04:40.250022+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.25820","citing_title":"Visual-Redundancy-Controlled Parallel Decoding for Diffusion-Based Multimodal Large Language Models","ref_index":11,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP","json":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP.json","graph_json":"https://pith.science/api/pith-number/K2BTAWY7XEM4DSRLAYPH5T6WAP/graph.json","events_json":"https://pith.science/api/pith-number/K2BTAWY7XEM4DSRLAYPH5T6WAP/events.json","paper":"https://pith.science/paper/K2BTAWY7"},"agent_actions":{"view_html":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP","download_json":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP.json","view_paper":"https://pith.science/paper/K2BTAWY7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.12996&json=true","fetch_graph":"https://pith.science/api/pith-number/K2BTAWY7XEM4DSRLAYPH5T6WAP/graph.json","fetch_events":"https://pith.science/api/pith-number/K2BTAWY7XEM4DSRLAYPH5T6WAP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP/action/storage_attestation","attest_author":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP/action/author_attestation","sign_citation":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP/action/citation_signature","submit_replication":"https://pith.science/pith/K2BTAWY7XEM4DSRLAYPH5T6WAP/action/replication_record"}},"created_at":"2026-06-02T03:04:40.250022+00:00","updated_at":"2026-06-02T03:04:40.250022+00:00"}