{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZG37Q2U4OBAOISEK2B6DESJOE7","short_pith_number":"pith:ZG37Q2U4","schema_version":"1.0","canonical_sha256":"c9b7f86a9c7040e4488ad07c32492e27fbaa3ac68c6ed684c751b1d8c5b9254c","source":{"kind":"arxiv","id":"2511.15098","version":2},"attestation_state":"computed","paper":{"title":"A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Duo Li, Ling Shao, Shijian Lu, Xiaoqin Zhang, Zuhao Yang","submitted_at":"2025-11-19T04:13:36Z","abstract_excerpt":"Discrete diffusion-based multimodal large language models (dMLLMs) have emerged as a promising alternative to autoregressive MLLMs thanks to their advantages in parallel decoding and bidirectional context modeling, but most existing dMLLMs incur significant computational overhead during inference due to the full-sequence attention computation in each denoising step. Pioneering studies attempt to resolve this issue from a modality-agnostic perspective via key-value cache optimization or efficient sampling but most of them overlook modality-specific visual token redundancy. In this work, we cond"},"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":"2511.15098","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-11-19T04:13:36Z","cross_cats_sorted":[],"title_canon_sha256":"6d22ee1291e1583dfc065018530a96f9ea2eb0c97f461ce72f4af5c225252093","abstract_canon_sha256":"cbd97e720b869f9dd1ffb1c3b87efd4704a1b724933e88666ef4a7ab8ff8713b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T02:13:18.485678Z","signature_b64":"JwzU8TvlP0VzvOMH5eguYSEvSeXMvrSY74qOmKsrvV9qVcqTtL3dW+B4WwBc9XNcqWm2s87kD6cJpD/DHO+qBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c9b7f86a9c7040e4488ad07c32492e27fbaa3ac68c6ed684c751b1d8c5b9254c","last_reissued_at":"2026-06-23T02:13:18.485204Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T02:13:18.485204Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Comprehensive Study on Visual Token Redundancy for Discrete Diffusion-based Multimodal Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Duo Li, Ling Shao, Shijian Lu, Xiaoqin Zhang, Zuhao Yang","submitted_at":"2025-11-19T04:13:36Z","abstract_excerpt":"Discrete diffusion-based multimodal large language models (dMLLMs) have emerged as a promising alternative to autoregressive MLLMs thanks to their advantages in parallel decoding and bidirectional context modeling, but most existing dMLLMs incur significant computational overhead during inference due to the full-sequence attention computation in each denoising step. Pioneering studies attempt to resolve this issue from a modality-agnostic perspective via key-value cache optimization or efficient sampling but most of them overlook modality-specific visual token redundancy. In this work, we cond"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.15098","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/2511.15098/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":"2511.15098","created_at":"2026-06-23T02:13:18.485261+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.15098v2","created_at":"2026-06-23T02:13:18.485261+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.15098","created_at":"2026-06-23T02:13:18.485261+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZG37Q2U4OBAO","created_at":"2026-06-23T02:13:18.485261+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZG37Q2U4OBAOISEK","created_at":"2026-06-23T02:13:18.485261+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZG37Q2U4","created_at":"2026-06-23T02:13:18.485261+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.25820","citing_title":"Visual-Redundancy-Controlled Parallel Decoding for Diffusion-Based Multimodal Large Language Models","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2604.05079","citing_title":"SVAgent: Storyline-Guided Long Video Understanding via Cross-Modal Multi-Agent Collaboration","ref_index":12,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7","json":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7.json","graph_json":"https://pith.science/api/pith-number/ZG37Q2U4OBAOISEK2B6DESJOE7/graph.json","events_json":"https://pith.science/api/pith-number/ZG37Q2U4OBAOISEK2B6DESJOE7/events.json","paper":"https://pith.science/paper/ZG37Q2U4"},"agent_actions":{"view_html":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7","download_json":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7.json","view_paper":"https://pith.science/paper/ZG37Q2U4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.15098&json=true","fetch_graph":"https://pith.science/api/pith-number/ZG37Q2U4OBAOISEK2B6DESJOE7/graph.json","fetch_events":"https://pith.science/api/pith-number/ZG37Q2U4OBAOISEK2B6DESJOE7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7/action/storage_attestation","attest_author":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7/action/author_attestation","sign_citation":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7/action/citation_signature","submit_replication":"https://pith.science/pith/ZG37Q2U4OBAOISEK2B6DESJOE7/action/replication_record"}},"created_at":"2026-06-23T02:13:18.485261+00:00","updated_at":"2026-06-23T02:13:18.485261+00:00"}