{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:HIBJPX2ARFUDEJCLY4KIMXCU77","short_pith_number":"pith:HIBJPX2A","schema_version":"1.0","canonical_sha256":"3a0297df40896832244bc714865c54fff8aa0207b0a5e71c685527846894471e","source":{"kind":"arxiv","id":"2501.00530","version":2},"attestation_state":"computed","paper":{"title":"Superposition in Transformers: A Novel Way of Building Mixture of Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ayoub Ben Chaliah, Hela Dellagi","submitted_at":"2024-12-31T16:28:23Z","abstract_excerpt":"Catastrophic forgetting remains a major challenge when adapting large language models (LLMs) to new tasks or domains. Conventional fine-tuning often overwrites existing knowledge, causing performance degradation on original tasks. We introduce Superposition in Transformers, a novel architecture that leverages autoencoders to superimpose the hidden representations of a base model and a fine-tuned model within a shared parameter space. By using B-spline-based blending coefficients and autoencoders that adaptively reconstruct hidden states based on the input data distribution, our method effectiv"},"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":"2501.00530","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-31T16:28:23Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ed11e15ffa320a3a0059fa8f79d15b68593f43478597f9028f2f0b4ff303dc4e","abstract_canon_sha256":"349546146844a39f7e4e9903161075d99412d6f2b249877c7625a846dbab6819"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:57:50.041296Z","signature_b64":"laogWXw2a8I0nJzKjWE6QJk9NBK4z9HuBHUfuEFgIwfcppL4+qNYfWmaN1SMfSCv2LLHFvOGbYGDtmXq9OxpCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3a0297df40896832244bc714865c54fff8aa0207b0a5e71c685527846894471e","last_reissued_at":"2026-07-05T09:57:50.040732Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:57:50.040732Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Superposition in Transformers: A Novel Way of Building Mixture of Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ayoub Ben Chaliah, Hela Dellagi","submitted_at":"2024-12-31T16:28:23Z","abstract_excerpt":"Catastrophic forgetting remains a major challenge when adapting large language models (LLMs) to new tasks or domains. Conventional fine-tuning often overwrites existing knowledge, causing performance degradation on original tasks. We introduce Superposition in Transformers, a novel architecture that leverages autoencoders to superimpose the hidden representations of a base model and a fine-tuned model within a shared parameter space. By using B-spline-based blending coefficients and autoencoders that adaptively reconstruct hidden states based on the input data distribution, our method effectiv"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.00530","kind":"arxiv","version":2},"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/2501.00530/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":"2501.00530","created_at":"2026-07-05T09:57:50.040790+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.00530v2","created_at":"2026-07-05T09:57:50.040790+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.00530","created_at":"2026-07-05T09:57:50.040790+00:00"},{"alias_kind":"pith_short_12","alias_value":"HIBJPX2ARFUD","created_at":"2026-07-05T09:57:50.040790+00:00"},{"alias_kind":"pith_short_16","alias_value":"HIBJPX2ARFUDEJCL","created_at":"2026-07-05T09:57:50.040790+00:00"},{"alias_kind":"pith_short_8","alias_value":"HIBJPX2A","created_at":"2026-07-05T09:57:50.040790+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/HIBJPX2ARFUDEJCLY4KIMXCU77","json":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77.json","graph_json":"https://pith.science/api/pith-number/HIBJPX2ARFUDEJCLY4KIMXCU77/graph.json","events_json":"https://pith.science/api/pith-number/HIBJPX2ARFUDEJCLY4KIMXCU77/events.json","paper":"https://pith.science/paper/HIBJPX2A"},"agent_actions":{"view_html":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77","download_json":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77.json","view_paper":"https://pith.science/paper/HIBJPX2A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.00530&json=true","fetch_graph":"https://pith.science/api/pith-number/HIBJPX2ARFUDEJCLY4KIMXCU77/graph.json","fetch_events":"https://pith.science/api/pith-number/HIBJPX2ARFUDEJCLY4KIMXCU77/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77/action/storage_attestation","attest_author":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77/action/author_attestation","sign_citation":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77/action/citation_signature","submit_replication":"https://pith.science/pith/HIBJPX2ARFUDEJCLY4KIMXCU77/action/replication_record"}},"created_at":"2026-07-05T09:57:50.040790+00:00","updated_at":"2026-07-05T09:57:50.040790+00:00"}