{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:76E5SMQRVE5WTTQNVTHY32CU2R","short_pith_number":"pith:76E5SMQR","schema_version":"1.0","canonical_sha256":"ff89d93211a93b69ce0daccf8de854d45f18d2ae2e0e3e46dfe0b7fcc66dcc38","source":{"kind":"arxiv","id":"2407.00875","version":1},"attestation_state":"computed","paper":{"title":"MoE-CT: A Novel Approach For Large Language Models Training With Resistance To Catastrophic Forgetting","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Baosong Yang, Binbin Xie, Deyi Xiong, Shangjie Li, Tianhao Li","submitted_at":"2024-06-25T11:03:45Z","abstract_excerpt":"The advent of large language models (LLMs) has predominantly catered to high-resource languages, leaving a disparity in performance for low-resource languages. Conventional Continual Training (CT) approaches to bridge this gap often undermine a model's original linguistic proficiency when expanding to multilingual contexts. Addressing this issue, we introduce a novel MoE-CT architecture, a paradigm that innovatively separates the base model's learning from the multilingual expansion process. Our design freezes the original LLM parameters, thus safeguarding its performance in high-resource lang"},"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":"2407.00875","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-25T11:03:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1f61ff5b1244a70cca418a4311539ab3c1372667792330eeea12ca3daed0163a","abstract_canon_sha256":"13c859049bdf99ddeff99794477ef66b98002e1ca12845ca0f5d38a94f64de41"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:38:25.804461Z","signature_b64":"/+zqIACwByWPCgxBahCUq7qGpQ4MZxPfQOEkd5G+8z79TlJIyXS4RvIjqtbt7fdKYE17/vwyPCBNbQC+TJw0AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff89d93211a93b69ce0daccf8de854d45f18d2ae2e0e3e46dfe0b7fcc66dcc38","last_reissued_at":"2026-07-05T08:38:25.804104Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:38:25.804104Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MoE-CT: A Novel Approach For Large Language Models Training With Resistance To Catastrophic Forgetting","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Baosong Yang, Binbin Xie, Deyi Xiong, Shangjie Li, Tianhao Li","submitted_at":"2024-06-25T11:03:45Z","abstract_excerpt":"The advent of large language models (LLMs) has predominantly catered to high-resource languages, leaving a disparity in performance for low-resource languages. Conventional Continual Training (CT) approaches to bridge this gap often undermine a model's original linguistic proficiency when expanding to multilingual contexts. Addressing this issue, we introduce a novel MoE-CT architecture, a paradigm that innovatively separates the base model's learning from the multilingual expansion process. Our design freezes the original LLM parameters, thus safeguarding its performance in high-resource lang"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.00875","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/2407.00875/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":"2407.00875","created_at":"2026-07-05T08:38:25.804160+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.00875v1","created_at":"2026-07-05T08:38:25.804160+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.00875","created_at":"2026-07-05T08:38:25.804160+00:00"},{"alias_kind":"pith_short_12","alias_value":"76E5SMQRVE5W","created_at":"2026-07-05T08:38:25.804160+00:00"},{"alias_kind":"pith_short_16","alias_value":"76E5SMQRVE5WTTQN","created_at":"2026-07-05T08:38:25.804160+00:00"},{"alias_kind":"pith_short_8","alias_value":"76E5SMQR","created_at":"2026-07-05T08:38:25.804160+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2509.03934","citing_title":"SelfAug: Mitigating Catastrophic Forgetting in Retrieval-Augmented Generation via Distribution Self-Alignment","ref_index":43,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R","json":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R.json","graph_json":"https://pith.science/api/pith-number/76E5SMQRVE5WTTQNVTHY32CU2R/graph.json","events_json":"https://pith.science/api/pith-number/76E5SMQRVE5WTTQNVTHY32CU2R/events.json","paper":"https://pith.science/paper/76E5SMQR"},"agent_actions":{"view_html":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R","download_json":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R.json","view_paper":"https://pith.science/paper/76E5SMQR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.00875&json=true","fetch_graph":"https://pith.science/api/pith-number/76E5SMQRVE5WTTQNVTHY32CU2R/graph.json","fetch_events":"https://pith.science/api/pith-number/76E5SMQRVE5WTTQNVTHY32CU2R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R/action/storage_attestation","attest_author":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R/action/author_attestation","sign_citation":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R/action/citation_signature","submit_replication":"https://pith.science/pith/76E5SMQRVE5WTTQNVTHY32CU2R/action/replication_record"}},"created_at":"2026-07-05T08:38:25.804160+00:00","updated_at":"2026-07-05T08:38:25.804160+00:00"}