{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FKWYWLBK2EIVLX3XWUTNJQ32GA","short_pith_number":"pith:FKWYWLBK","schema_version":"1.0","canonical_sha256":"2aad8b2c2ad11155df77b526d4c37a302c90a5c868b288c7a4b9caedf4a76abd","source":{"kind":"arxiv","id":"2607.14557","version":1},"attestation_state":"computed","paper":{"title":"Seeing the End at Step Zero: Accelerating Diffusion MLLMs via MLP Sparsity-Aware Truncation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Qicheng Zhao, Qi Sun, Zheyu Yan","submitted_at":"2026-07-16T04:37:02Z","abstract_excerpt":"Diffusion Multimodal Large Language Models (DMLLMs) are highly effective for multimodal reasoning, yet their inference efficiency is significantly hindered by fixed-length generation constraints. Since the actual output length is unknown, output sequences are padded to a predefined maximum length, resulting in substantial redundant computation over unnecessary [EOS] tokens. In this work, we discover that DMLLMs implicitly reveal their valid semantic boundary at the very first denoising step through a distinct shift in MLP activation sparsity. Leveraging this observation, we propose Seer, a tra"},"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":"2607.14557","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-16T04:37:02Z","cross_cats_sorted":[],"title_canon_sha256":"9fef5f239e4ccfdd032636dbb28214cc2d0cb15c4d63c0562ad8987af410e390","abstract_canon_sha256":"720fb035c879477d9d972e9c58a92c2893e34fcfbe8ef81c13fe9aeaade3a866"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-17T01:21:18.004324Z","signature_b64":"16824upiLlNGHeWchiD563LUG7rIuiWZLlh48rzzUBiNNQlBGrERbQvk14ibjuTOArryFKO0AwWB0Jf1WAGnDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2aad8b2c2ad11155df77b526d4c37a302c90a5c868b288c7a4b9caedf4a76abd","last_reissued_at":"2026-07-17T01:21:18.003491Z","signature_status":"signed_v1","first_computed_at":"2026-07-17T01:21:18.003491Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Seeing the End at Step Zero: Accelerating Diffusion MLLMs via MLP Sparsity-Aware Truncation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Qicheng Zhao, Qi Sun, Zheyu Yan","submitted_at":"2026-07-16T04:37:02Z","abstract_excerpt":"Diffusion Multimodal Large Language Models (DMLLMs) are highly effective for multimodal reasoning, yet their inference efficiency is significantly hindered by fixed-length generation constraints. Since the actual output length is unknown, output sequences are padded to a predefined maximum length, resulting in substantial redundant computation over unnecessary [EOS] tokens. In this work, we discover that DMLLMs implicitly reveal their valid semantic boundary at the very first denoising step through a distinct shift in MLP activation sparsity. Leveraging this observation, we propose Seer, a tra"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.14557","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/2607.14557/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":"2607.14557","created_at":"2026-07-17T01:21:18.003916+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.14557v1","created_at":"2026-07-17T01:21:18.003916+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.14557","created_at":"2026-07-17T01:21:18.003916+00:00"},{"alias_kind":"pith_short_12","alias_value":"FKWYWLBK2EIV","created_at":"2026-07-17T01:21:18.003916+00:00"},{"alias_kind":"pith_short_16","alias_value":"FKWYWLBK2EIVLX3X","created_at":"2026-07-17T01:21:18.003916+00:00"},{"alias_kind":"pith_short_8","alias_value":"FKWYWLBK","created_at":"2026-07-17T01:21:18.003916+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/FKWYWLBK2EIVLX3XWUTNJQ32GA","json":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA.json","graph_json":"https://pith.science/api/pith-number/FKWYWLBK2EIVLX3XWUTNJQ32GA/graph.json","events_json":"https://pith.science/api/pith-number/FKWYWLBK2EIVLX3XWUTNJQ32GA/events.json","paper":"https://pith.science/paper/FKWYWLBK"},"agent_actions":{"view_html":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA","download_json":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA.json","view_paper":"https://pith.science/paper/FKWYWLBK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.14557&json=true","fetch_graph":"https://pith.science/api/pith-number/FKWYWLBK2EIVLX3XWUTNJQ32GA/graph.json","fetch_events":"https://pith.science/api/pith-number/FKWYWLBK2EIVLX3XWUTNJQ32GA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA/action/storage_attestation","attest_author":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA/action/author_attestation","sign_citation":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA/action/citation_signature","submit_replication":"https://pith.science/pith/FKWYWLBK2EIVLX3XWUTNJQ32GA/action/replication_record"}},"created_at":"2026-07-17T01:21:18.003916+00:00","updated_at":"2026-07-17T01:21:18.003916+00:00"}