{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:WDCIBQBTXOMJAL6Y4NVMOTEVBL","short_pith_number":"pith:WDCIBQBT","schema_version":"1.0","canonical_sha256":"b0c480c033bb98902fd8e36ac74c950afc939ee7bfefca2c1e240604d31a025a","source":{"kind":"arxiv","id":"2606.18304","version":1},"attestation_state":"computed","paper":{"title":"Attribution-Guided and Coverage-Maximized Pruning for Structural MoE Compression","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dacheng Tao, Ge Yang, Jiacheng Wang, Jinyang Guo, Xianglong Liu, Yifu Ding, Yongcheng Jing","submitted_at":"2026-06-16T06:53:27Z","abstract_excerpt":"Mixture-of-Experts (MoE) models scale compute efficiently, yet remain expensive to deploy due to their substantial memory footprint and inference overhead. Prior compression methods mainly operate at the expert level, either removing entire experts or ranking experts by coarse-grained importance scores. However, such expert-wise decisions are often too coarse to capture fine-grained redundancy, leading to misallocated pruning budgets and limited compression. To address this problem, we observe that information within MoE experts is highly concentrated in a small subset of channels, leaving sub"},"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":"2606.18304","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-16T06:53:27Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"f644d9a59dd036366736c5dbbe2e20021eae341d9ae8e61eba96b7eefd56ce28","abstract_canon_sha256":"cdb37b4c438372cf2b7cc49870c266a4228a9da389f98700779a552103668175"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:10:57.447376Z","signature_b64":"oQDwQXXhRkeMesGl4UwuOrElOYtQrcooggUB8s14+CfiesKuzh20rirtOo6MTLZ+wBA07pGEvMK3csOl2Zw0Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b0c480c033bb98902fd8e36ac74c950afc939ee7bfefca2c1e240604d31a025a","last_reissued_at":"2026-06-19T16:10:57.447021Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:10:57.447021Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Attribution-Guided and Coverage-Maximized Pruning for Structural MoE Compression","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dacheng Tao, Ge Yang, Jiacheng Wang, Jinyang Guo, Xianglong Liu, Yifu Ding, Yongcheng Jing","submitted_at":"2026-06-16T06:53:27Z","abstract_excerpt":"Mixture-of-Experts (MoE) models scale compute efficiently, yet remain expensive to deploy due to their substantial memory footprint and inference overhead. Prior compression methods mainly operate at the expert level, either removing entire experts or ranking experts by coarse-grained importance scores. However, such expert-wise decisions are often too coarse to capture fine-grained redundancy, leading to misallocated pruning budgets and limited compression. To address this problem, we observe that information within MoE experts is highly concentrated in a small subset of channels, leaving sub"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.18304","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/2606.18304/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":"2606.18304","created_at":"2026-06-19T16:10:57.447081+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.18304v1","created_at":"2026-06-19T16:10:57.447081+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.18304","created_at":"2026-06-19T16:10:57.447081+00:00"},{"alias_kind":"pith_short_12","alias_value":"WDCIBQBTXOMJ","created_at":"2026-06-19T16:10:57.447081+00:00"},{"alias_kind":"pith_short_16","alias_value":"WDCIBQBTXOMJAL6Y","created_at":"2026-06-19T16:10:57.447081+00:00"},{"alias_kind":"pith_short_8","alias_value":"WDCIBQBT","created_at":"2026-06-19T16:10:57.447081+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/WDCIBQBTXOMJAL6Y4NVMOTEVBL","json":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL.json","graph_json":"https://pith.science/api/pith-number/WDCIBQBTXOMJAL6Y4NVMOTEVBL/graph.json","events_json":"https://pith.science/api/pith-number/WDCIBQBTXOMJAL6Y4NVMOTEVBL/events.json","paper":"https://pith.science/paper/WDCIBQBT"},"agent_actions":{"view_html":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL","download_json":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL.json","view_paper":"https://pith.science/paper/WDCIBQBT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.18304&json=true","fetch_graph":"https://pith.science/api/pith-number/WDCIBQBTXOMJAL6Y4NVMOTEVBL/graph.json","fetch_events":"https://pith.science/api/pith-number/WDCIBQBTXOMJAL6Y4NVMOTEVBL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL/action/storage_attestation","attest_author":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL/action/author_attestation","sign_citation":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL/action/citation_signature","submit_replication":"https://pith.science/pith/WDCIBQBTXOMJAL6Y4NVMOTEVBL/action/replication_record"}},"created_at":"2026-06-19T16:10:57.447081+00:00","updated_at":"2026-06-19T16:10:57.447081+00:00"}