{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:NOQ2CXTHH2M63NQKVSYA3EUCGS","short_pith_number":"pith:NOQ2CXTH","schema_version":"1.0","canonical_sha256":"6ba1a15e673e99edb60aacb00d9282349fd297502d91c3859acbe1254b142da2","source":{"kind":"arxiv","id":"2506.09105","version":3},"attestation_state":"computed","paper":{"title":"MetaTT: A Global Tensor-Train Adapter for Parameter-Efficient Fine-Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","quant-ph"],"primary_cat":"cs.LG","authors_text":"Archan Ray, Javier Lopez-Piqueres, Marco Pistoia, Mattia J. Villani, Niraj Kumar, Pranav Deshpande","submitted_at":"2025-06-10T16:32:05Z","abstract_excerpt":"We present MetaTT, a Tensor Train (TT) adapter framework for fine-tuning of pre-trained transformers. MetaTT enables flexible and parameter-efficient model adaptation by using a single shared TT to factorize transformer sub-modules. This factorization indexes key structural dimensions, including layer and matrix type, and can optionally incorporate heads and tasks. This design allows MetaTT's parameter count to scale with the sum, rather than the product, of the modes, resulting in a substantially more compact adapter. Our benchmarks compare MetaTT with LoRA along with recent state-of-the-art "},"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":"2506.09105","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-10T16:32:05Z","cross_cats_sorted":["cs.AI","quant-ph"],"title_canon_sha256":"498dcadac6ac647ba8c28717cae81a4184271df8d86adbf1434356e86bd693ed","abstract_canon_sha256":"461d9511c68bac37b9fec3181b4dd14ea55fee718576cb6548fc1b4b706ff43b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-03T01:17:11.805431Z","signature_b64":"2ZimjIfps1Sfqlm9aB3YrbapdBCZsjUVls/MfCDHLzsTGRPQdkwUa6PrBmDP0OjVPa5LIueYacAhvrR8qu3TDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6ba1a15e673e99edb60aacb00d9282349fd297502d91c3859acbe1254b142da2","last_reissued_at":"2026-07-03T01:17:11.804905Z","signature_status":"signed_v1","first_computed_at":"2026-07-03T01:17:11.804905Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MetaTT: A Global Tensor-Train Adapter for Parameter-Efficient Fine-Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","quant-ph"],"primary_cat":"cs.LG","authors_text":"Archan Ray, Javier Lopez-Piqueres, Marco Pistoia, Mattia J. Villani, Niraj Kumar, Pranav Deshpande","submitted_at":"2025-06-10T16:32:05Z","abstract_excerpt":"We present MetaTT, a Tensor Train (TT) adapter framework for fine-tuning of pre-trained transformers. MetaTT enables flexible and parameter-efficient model adaptation by using a single shared TT to factorize transformer sub-modules. This factorization indexes key structural dimensions, including layer and matrix type, and can optionally incorporate heads and tasks. This design allows MetaTT's parameter count to scale with the sum, rather than the product, of the modes, resulting in a substantially more compact adapter. Our benchmarks compare MetaTT with LoRA along with recent state-of-the-art "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.09105","kind":"arxiv","version":3},"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/2506.09105/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":"2506.09105","created_at":"2026-07-03T01:17:11.804965+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.09105v3","created_at":"2026-07-03T01:17:11.804965+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.09105","created_at":"2026-07-03T01:17:11.804965+00:00"},{"alias_kind":"pith_short_12","alias_value":"NOQ2CXTHH2M6","created_at":"2026-07-03T01:17:11.804965+00:00"},{"alias_kind":"pith_short_16","alias_value":"NOQ2CXTHH2M63NQK","created_at":"2026-07-03T01:17:11.804965+00:00"},{"alias_kind":"pith_short_8","alias_value":"NOQ2CXTH","created_at":"2026-07-03T01:17:11.804965+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.16769","citing_title":"GLT-PEFT: Gated Lie-Tucker Parameter-Efficient Fine-Tuning for Alzheimer's Disease Diagnosis with Hippocampal Segmentation Pretraining","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2604.21905","citing_title":"Low-Rank Adaptation Redux for Large Models","ref_index":120,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS","json":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS.json","graph_json":"https://pith.science/api/pith-number/NOQ2CXTHH2M63NQKVSYA3EUCGS/graph.json","events_json":"https://pith.science/api/pith-number/NOQ2CXTHH2M63NQKVSYA3EUCGS/events.json","paper":"https://pith.science/paper/NOQ2CXTH"},"agent_actions":{"view_html":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS","download_json":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS.json","view_paper":"https://pith.science/paper/NOQ2CXTH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.09105&json=true","fetch_graph":"https://pith.science/api/pith-number/NOQ2CXTHH2M63NQKVSYA3EUCGS/graph.json","fetch_events":"https://pith.science/api/pith-number/NOQ2CXTHH2M63NQKVSYA3EUCGS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS/action/storage_attestation","attest_author":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS/action/author_attestation","sign_citation":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS/action/citation_signature","submit_replication":"https://pith.science/pith/NOQ2CXTHH2M63NQKVSYA3EUCGS/action/replication_record"}},"created_at":"2026-07-03T01:17:11.804965+00:00","updated_at":"2026-07-03T01:17:11.804965+00:00"}