{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ZSDM76VOSPBMS5VQ4VEJ4JMHAP","short_pith_number":"pith:ZSDM76VO","schema_version":"1.0","canonical_sha256":"cc86cffaae93c2c976b0e5489e258703e6bee15f58a528374ddf4ae6038bd6cf","source":{"kind":"arxiv","id":"2406.16554","version":1},"attestation_state":"computed","paper":{"title":"LLaMA-MoE: Building Mixture-of-Experts from LLaMA with Continual Pre-training","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Conghui He, Daize Dong, Jiacheng Ruan, Jingqi Tong, Tong Zhu, Xiaoye Qu, Yu Cheng","submitted_at":"2024-06-24T11:43:07Z","abstract_excerpt":"Mixture-of-Experts (MoE) has gained increasing popularity as a promising framework for scaling up large language models (LLMs). However, training MoE from scratch in a large-scale setting still suffers from data-hungry and instability problems. Motivated by this limit, we investigate building MoE models from existing dense large language models. Specifically, based on the well-known LLaMA-2 7B model, we obtain an MoE model by: (1) Expert Construction, which partitions the parameters of original Feed-Forward Networks (FFNs) into multiple experts; (2) Continual Pre-training, which further trains"},"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":"2406.16554","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-24T11:43:07Z","cross_cats_sorted":[],"title_canon_sha256":"561737ad7805091c075f7a11e812c051b0a249fb745514a7a7090f0138c4707e","abstract_canon_sha256":"9b284cdca4e8266b4403448127055686f491d0097a7d52a65107f3f71e614444"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:36:00.418901Z","signature_b64":"zckgotPRCA6ttnJkFvxgxEZe981Wxz4a/0V3NuWi4MeSbDGWXiYi8tWVLlzxPP/1NSCLp4vmY7Zrl99HYllNCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cc86cffaae93c2c976b0e5489e258703e6bee15f58a528374ddf4ae6038bd6cf","last_reissued_at":"2026-07-05T08:36:00.418488Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:36:00.418488Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLaMA-MoE: Building Mixture-of-Experts from LLaMA with Continual Pre-training","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Conghui He, Daize Dong, Jiacheng Ruan, Jingqi Tong, Tong Zhu, Xiaoye Qu, Yu Cheng","submitted_at":"2024-06-24T11:43:07Z","abstract_excerpt":"Mixture-of-Experts (MoE) has gained increasing popularity as a promising framework for scaling up large language models (LLMs). However, training MoE from scratch in a large-scale setting still suffers from data-hungry and instability problems. Motivated by this limit, we investigate building MoE models from existing dense large language models. Specifically, based on the well-known LLaMA-2 7B model, we obtain an MoE model by: (1) Expert Construction, which partitions the parameters of original Feed-Forward Networks (FFNs) into multiple experts; (2) Continual Pre-training, which further trains"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.16554","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/2406.16554/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":"2406.16554","created_at":"2026-07-05T08:36:00.418543+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.16554v1","created_at":"2026-07-05T08:36:00.418543+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.16554","created_at":"2026-07-05T08:36:00.418543+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZSDM76VOSPBM","created_at":"2026-07-05T08:36:00.418543+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZSDM76VOSPBMS5VQ","created_at":"2026-07-05T08:36:00.418543+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZSDM76VO","created_at":"2026-07-05T08:36:00.418543+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08814","citing_title":"STAR: Rethinking MoE Routing as Structure-Aware Subspace Learning","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2503.05066","citing_title":"Capacity-Aware Inference: Mitigating the Straggler Effect in Mixture of Experts","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00515","citing_title":"SpaceMoE: Realizing Distributed Mixture-of-Experts Inference over Space Networks","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2509.08470","citing_title":"Joint Learning using Mixture-of-Expert-Based Representation for Speech Enhancement and Robust Emotion Recognition","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08292","citing_title":"Hierarchical Mixture-of-Experts with Two-Stage Optimization","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00515","citing_title":"SpaceMoE: Realizing Distributed Mixture-of-Experts Inference over Space Networks","ref_index":41,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP","json":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP.json","graph_json":"https://pith.science/api/pith-number/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/graph.json","events_json":"https://pith.science/api/pith-number/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/events.json","paper":"https://pith.science/paper/ZSDM76VO"},"agent_actions":{"view_html":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP","download_json":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP.json","view_paper":"https://pith.science/paper/ZSDM76VO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.16554&json=true","fetch_graph":"https://pith.science/api/pith-number/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/graph.json","fetch_events":"https://pith.science/api/pith-number/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/action/storage_attestation","attest_author":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/action/author_attestation","sign_citation":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/action/citation_signature","submit_replication":"https://pith.science/pith/ZSDM76VOSPBMS5VQ4VEJ4JMHAP/action/replication_record"}},"created_at":"2026-07-05T08:36:00.418543+00:00","updated_at":"2026-07-05T08:36:00.418543+00:00"}