{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:F3X6ZENMBD3K422MOOLDZA2BU5","short_pith_number":"pith:F3X6ZENM","schema_version":"1.0","canonical_sha256":"2eefec91ac08f6ae6b4c73963c8341a74f8ca59f455045db71a9c088a9236e44","source":{"kind":"arxiv","id":"2203.01104","version":4},"attestation_state":"computed","paper":{"title":"Parameter-Efficient Mixture-of-Experts Architecture for Pre-trained Language Models","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.AI","cs.LG","quant-ph"],"primary_cat":"cs.CL","authors_text":"Ji-Rong Wen, Peiyu Liu, Wayne Xin Zhao, Ze-Feng Gao, Zhong-Yi Lu","submitted_at":"2022-03-02T13:44:49Z","abstract_excerpt":"Recently, Mixture-of-Experts (short as MoE) architecture has achieved remarkable success in increasing the model capacity of large-scale language models. However, MoE requires incorporating significantly more parameters than the base model being extended. In this paper, we propose building a parameter-efficient MoE architecture by sharing information among experts. We adopt the matrix product operator (MPO, a tensor decomposition from quantum many-body physics) to reconstruct the parameter matrix in the expert layer and increase model capacity for pre-trained language models by sharing paramet"},"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":"2203.01104","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-02T13:44:49Z","cross_cats_sorted":["cs.AI","cs.LG","quant-ph"],"title_canon_sha256":"ccd1b05d83ef0bc57cb79ea23d6f37822b90d4e7c076323a65d40f01440fb6d7","abstract_canon_sha256":"b650a10c3df1ec264c5fecb45a31c48dd508e22ce0ea4504449ac54f8308ac15"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:06:10.270743Z","signature_b64":"pkNU+BdTId5jdGvmB26ioqAZwp9LGig0HAqnlKUi7K8tDFHfoWuu3g+zVJkZq5midGCf3ha8J8Fpb/Or9YzYDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2eefec91ac08f6ae6b4c73963c8341a74f8ca59f455045db71a9c088a9236e44","last_reissued_at":"2026-07-05T05:06:10.270176Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:06:10.270176Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Parameter-Efficient Mixture-of-Experts Architecture for Pre-trained Language Models","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.AI","cs.LG","quant-ph"],"primary_cat":"cs.CL","authors_text":"Ji-Rong Wen, Peiyu Liu, Wayne Xin Zhao, Ze-Feng Gao, Zhong-Yi Lu","submitted_at":"2022-03-02T13:44:49Z","abstract_excerpt":"Recently, Mixture-of-Experts (short as MoE) architecture has achieved remarkable success in increasing the model capacity of large-scale language models. However, MoE requires incorporating significantly more parameters than the base model being extended. In this paper, we propose building a parameter-efficient MoE architecture by sharing information among experts. We adopt the matrix product operator (MPO, a tensor decomposition from quantum many-body physics) to reconstruct the parameter matrix in the expert layer and increase model capacity for pre-trained language models by sharing paramet"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.01104","kind":"arxiv","version":4},"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/2203.01104/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":"2203.01104","created_at":"2026-07-05T05:06:10.270266+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.01104v4","created_at":"2026-07-05T05:06:10.270266+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.01104","created_at":"2026-07-05T05:06:10.270266+00:00"},{"alias_kind":"pith_short_12","alias_value":"F3X6ZENMBD3K","created_at":"2026-07-05T05:06:10.270266+00:00"},{"alias_kind":"pith_short_16","alias_value":"F3X6ZENMBD3K422M","created_at":"2026-07-05T05:06:10.270266+00:00"},{"alias_kind":"pith_short_8","alias_value":"F3X6ZENM","created_at":"2026-07-05T05:06:10.270266+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22541","citing_title":"ASAP: A Disaggregated and Asynchronous Inference System for MoE Prefill","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16470","citing_title":"Strategic Over-Parameterization for Generalizable Low-Rank Adaptation","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5","json":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5.json","graph_json":"https://pith.science/api/pith-number/F3X6ZENMBD3K422MOOLDZA2BU5/graph.json","events_json":"https://pith.science/api/pith-number/F3X6ZENMBD3K422MOOLDZA2BU5/events.json","paper":"https://pith.science/paper/F3X6ZENM"},"agent_actions":{"view_html":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5","download_json":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5.json","view_paper":"https://pith.science/paper/F3X6ZENM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.01104&json=true","fetch_graph":"https://pith.science/api/pith-number/F3X6ZENMBD3K422MOOLDZA2BU5/graph.json","fetch_events":"https://pith.science/api/pith-number/F3X6ZENMBD3K422MOOLDZA2BU5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5/action/storage_attestation","attest_author":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5/action/author_attestation","sign_citation":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5/action/citation_signature","submit_replication":"https://pith.science/pith/F3X6ZENMBD3K422MOOLDZA2BU5/action/replication_record"}},"created_at":"2026-07-05T05:06:10.270266+00:00","updated_at":"2026-07-05T05:06:10.270266+00:00"}