{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JAP4O74ZWYKIXYMQB5Y343OVBA","short_pith_number":"pith:JAP4O74Z","schema_version":"1.0","canonical_sha256":"481fc77f99b6148be1900f71be6dd508125bdee7b3ea7232b7bdf0ccfb1de835","source":{"kind":"arxiv","id":"2602.08404","version":2},"attestation_state":"computed","paper":{"title":"TEAM: Temporal-Spatial Consistency Guided Expert Activation for MoE Diffusion Language Model Acceleration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Linye Wei, Meng Li, Pingzhi Tang, Zixiang Luo","submitted_at":"2026-02-09T09:05:46Z","abstract_excerpt":"Diffusion large language models (dLLMs) have recently gained significant attention due to their inherent support for parallel decoding. Building on this paradigm, Mixture-of-Experts (MoE) dLLMs with autoregressive (AR) initialization have further demonstrated strong performance competitive with mainstream AR models. However, we identify a fundamental mismatch between MoE architectures and diffusion-based decoding. Specifically, a large number of experts are activated at each denoising step, while only a small subset of tokens is ultimately accepted, resulting in substantial inference overhead "},"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":"2602.08404","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-02-09T09:05:46Z","cross_cats_sorted":[],"title_canon_sha256":"77522b2cd3d8d266d07fc5011f46baebce180009f3b358944d8748164e2f72ba","abstract_canon_sha256":"41c095a082301e596f4581d490d32c3f1418aba7f96507b4dd636cae39abd34b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:02:13.238254Z","signature_b64":"ElWIePsPijuXj6XXD0+HH+2KBg0OEJwwQQhqI92i38fM3Kdhh8tpxwBIcAs9UX/cwO2lEPXpk4/f84wJI59jCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"481fc77f99b6148be1900f71be6dd508125bdee7b3ea7232b7bdf0ccfb1de835","last_reissued_at":"2026-05-25T02:02:13.237441Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:02:13.237441Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TEAM: Temporal-Spatial Consistency Guided Expert Activation for MoE Diffusion Language Model Acceleration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Linye Wei, Meng Li, Pingzhi Tang, Zixiang Luo","submitted_at":"2026-02-09T09:05:46Z","abstract_excerpt":"Diffusion large language models (dLLMs) have recently gained significant attention due to their inherent support for parallel decoding. Building on this paradigm, Mixture-of-Experts (MoE) dLLMs with autoregressive (AR) initialization have further demonstrated strong performance competitive with mainstream AR models. However, we identify a fundamental mismatch between MoE architectures and diffusion-based decoding. Specifically, a large number of experts are activated at each denoising step, while only a small subset of tokens is ultimately accepted, resulting in substantial inference overhead "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.08404","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/2602.08404/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":"2602.08404","created_at":"2026-05-25T02:02:13.237561+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.08404v2","created_at":"2026-05-25T02:02:13.237561+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.08404","created_at":"2026-05-25T02:02:13.237561+00:00"},{"alias_kind":"pith_short_12","alias_value":"JAP4O74ZWYKI","created_at":"2026-05-25T02:02:13.237561+00:00"},{"alias_kind":"pith_short_16","alias_value":"JAP4O74ZWYKIXYMQ","created_at":"2026-05-25T02:02:13.237561+00:00"},{"alias_kind":"pith_short_8","alias_value":"JAP4O74Z","created_at":"2026-05-25T02:02:13.237561+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.15855","citing_title":"Do Less, Achieve More: Do We Need Every-Step Optimization for RL Fine-tuning of Diffusion Models?","ref_index":55,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA","json":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA.json","graph_json":"https://pith.science/api/pith-number/JAP4O74ZWYKIXYMQB5Y343OVBA/graph.json","events_json":"https://pith.science/api/pith-number/JAP4O74ZWYKIXYMQB5Y343OVBA/events.json","paper":"https://pith.science/paper/JAP4O74Z"},"agent_actions":{"view_html":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA","download_json":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA.json","view_paper":"https://pith.science/paper/JAP4O74Z","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.08404&json=true","fetch_graph":"https://pith.science/api/pith-number/JAP4O74ZWYKIXYMQB5Y343OVBA/graph.json","fetch_events":"https://pith.science/api/pith-number/JAP4O74ZWYKIXYMQB5Y343OVBA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA/action/storage_attestation","attest_author":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA/action/author_attestation","sign_citation":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA/action/citation_signature","submit_replication":"https://pith.science/pith/JAP4O74ZWYKIXYMQB5Y343OVBA/action/replication_record"}},"created_at":"2026-05-25T02:02:13.237561+00:00","updated_at":"2026-05-25T02:02:13.237561+00:00"}