{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:34ORHNJD75IMPPXAPFXKQJ5LJR","short_pith_number":"pith:34ORHNJD","schema_version":"1.0","canonical_sha256":"df1d13b523ff50c7bee0796ea827ab4c45a62c6041068dd0f2af65f4bb16ee40","source":{"kind":"arxiv","id":"2506.09351","version":1},"attestation_state":"computed","paper":{"title":"DIVE into MoE: Diversity-Enhanced Reconstruction of Large Language Models from Dense into Mixture-of-Experts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bowen Shen, Jiaxuan Zhao, Naibin Gu, Peng Fu, Weiping Wang, Yuchen Feng, Zheng Lin","submitted_at":"2025-06-11T03:05:24Z","abstract_excerpt":"Large language models (LLMs) with the Mixture-of-Experts (MoE) architecture achieve high cost-efficiency by selectively activating a subset of the parameters. Despite the inference efficiency of MoE LLMs, the training of extensive experts from scratch incurs substantial overhead, whereas reconstructing a dense LLM into an MoE LLM significantly reduces the training budget. However, existing reconstruction methods often overlook the diversity among experts, leading to potential redundancy. In this paper, we come up with the observation that a specific LLM exhibits notable diversity after being p"},"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":"2506.09351","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-06-11T03:05:24Z","cross_cats_sorted":[],"title_canon_sha256":"80e5702345ca9cf0a9684d7d7ace55c28549a994b253c34d2e6f580213eefc10","abstract_canon_sha256":"5af2aa7f812f64653174bac26c967000f9055d56e3899fa0236643280aff5cdd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:19:35.470348Z","signature_b64":"1CwZL2Dq82lb4cDY/u0lMBySE8XGgj3mGOy3yA+/PiObciYsCLqCtxjR85FAaLMd5GInHTH4QF3MBBiWKgkTCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"df1d13b523ff50c7bee0796ea827ab4c45a62c6041068dd0f2af65f4bb16ee40","last_reissued_at":"2026-07-05T11:19:35.469872Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:19:35.469872Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DIVE into MoE: Diversity-Enhanced Reconstruction of Large Language Models from Dense into Mixture-of-Experts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bowen Shen, Jiaxuan Zhao, Naibin Gu, Peng Fu, Weiping Wang, Yuchen Feng, Zheng Lin","submitted_at":"2025-06-11T03:05:24Z","abstract_excerpt":"Large language models (LLMs) with the Mixture-of-Experts (MoE) architecture achieve high cost-efficiency by selectively activating a subset of the parameters. Despite the inference efficiency of MoE LLMs, the training of extensive experts from scratch incurs substantial overhead, whereas reconstructing a dense LLM into an MoE LLM significantly reduces the training budget. However, existing reconstruction methods often overlook the diversity among experts, leading to potential redundancy. In this paper, we come up with the observation that a specific LLM exhibits notable diversity after being p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.09351","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/2506.09351/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":"2506.09351","created_at":"2026-07-05T11:19:35.469927+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.09351v1","created_at":"2026-07-05T11:19:35.469927+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.09351","created_at":"2026-07-05T11:19:35.469927+00:00"},{"alias_kind":"pith_short_12","alias_value":"34ORHNJD75IM","created_at":"2026-07-05T11:19:35.469927+00:00"},{"alias_kind":"pith_short_16","alias_value":"34ORHNJD75IMPPXA","created_at":"2026-07-05T11:19:35.469927+00:00"},{"alias_kind":"pith_short_8","alias_value":"34ORHNJD","created_at":"2026-07-05T11:19:35.469927+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/34ORHNJD75IMPPXAPFXKQJ5LJR","json":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR.json","graph_json":"https://pith.science/api/pith-number/34ORHNJD75IMPPXAPFXKQJ5LJR/graph.json","events_json":"https://pith.science/api/pith-number/34ORHNJD75IMPPXAPFXKQJ5LJR/events.json","paper":"https://pith.science/paper/34ORHNJD"},"agent_actions":{"view_html":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR","download_json":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR.json","view_paper":"https://pith.science/paper/34ORHNJD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.09351&json=true","fetch_graph":"https://pith.science/api/pith-number/34ORHNJD75IMPPXAPFXKQJ5LJR/graph.json","fetch_events":"https://pith.science/api/pith-number/34ORHNJD75IMPPXAPFXKQJ5LJR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR/action/storage_attestation","attest_author":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR/action/author_attestation","sign_citation":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR/action/citation_signature","submit_replication":"https://pith.science/pith/34ORHNJD75IMPPXAPFXKQJ5LJR/action/replication_record"}},"created_at":"2026-07-05T11:19:35.469927+00:00","updated_at":"2026-07-05T11:19:35.469927+00:00"}