{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MBCF2YU64WYWT6FHFLBWF2EAAQ","short_pith_number":"pith:MBCF2YU6","schema_version":"1.0","canonical_sha256":"60445d629ee5b169f8a72ac362e8800415d73fb4d4dd1239068203dc183c3412","source":{"kind":"arxiv","id":"2507.08044","version":1},"attestation_state":"computed","paper":{"title":"ConsNoTrainLoRA: Data-driven Weight Initialization of Low-rank Adapters using Constraints","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Debasmit Das, Fatih Porikli, Hyoungwoo Park, Munawar Hayat, Seokeon Choi, Sungrack Yun","submitted_at":"2025-07-09T23:52:31Z","abstract_excerpt":"Foundation models are pre-trained on large-scale datasets and subsequently fine-tuned on small-scale datasets using parameter-efficient fine-tuning (PEFT) techniques like low-rank adapters (LoRA). In most previous works, LoRA weight matrices are randomly initialized with a fixed rank across all attachment points. In this paper, we improve convergence and final performance of LoRA fine-tuning, using our proposed data-driven weight initialization method, ConsNoTrainLoRA (CNTLoRA). We express LoRA initialization as a domain shift problem where we use multiple constraints relating the pre-training"},"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":"2507.08044","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-07-09T23:52:31Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2254cb7dc672aefac81cbec2d498165239f87995e380ed32ee13a567abb6236f","abstract_canon_sha256":"558bb6cd9c96919551e61c7f24568a279f94a19d2a2c48818c91f0e0f472fb67"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:35:16.274953Z","signature_b64":"Nx5orce08/QF3z0awJJNgXD1yBE6L79GxD+J9xl9Fh2KDRuWAbH+Y3654XXMzxRs/5WVIdQU1XUzAO+SSJNpBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"60445d629ee5b169f8a72ac362e8800415d73fb4d4dd1239068203dc183c3412","last_reissued_at":"2026-07-05T11:35:16.274441Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:35:16.274441Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ConsNoTrainLoRA: Data-driven Weight Initialization of Low-rank Adapters using Constraints","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Debasmit Das, Fatih Porikli, Hyoungwoo Park, Munawar Hayat, Seokeon Choi, Sungrack Yun","submitted_at":"2025-07-09T23:52:31Z","abstract_excerpt":"Foundation models are pre-trained on large-scale datasets and subsequently fine-tuned on small-scale datasets using parameter-efficient fine-tuning (PEFT) techniques like low-rank adapters (LoRA). In most previous works, LoRA weight matrices are randomly initialized with a fixed rank across all attachment points. In this paper, we improve convergence and final performance of LoRA fine-tuning, using our proposed data-driven weight initialization method, ConsNoTrainLoRA (CNTLoRA). We express LoRA initialization as a domain shift problem where we use multiple constraints relating the pre-training"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.08044","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/2507.08044/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":"2507.08044","created_at":"2026-07-05T11:35:16.274505+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.08044v1","created_at":"2026-07-05T11:35:16.274505+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.08044","created_at":"2026-07-05T11:35:16.274505+00:00"},{"alias_kind":"pith_short_12","alias_value":"MBCF2YU64WYW","created_at":"2026-07-05T11:35:16.274505+00:00"},{"alias_kind":"pith_short_16","alias_value":"MBCF2YU64WYWT6FH","created_at":"2026-07-05T11:35:16.274505+00:00"},{"alias_kind":"pith_short_8","alias_value":"MBCF2YU6","created_at":"2026-07-05T11:35:16.274505+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/MBCF2YU64WYWT6FHFLBWF2EAAQ","json":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ.json","graph_json":"https://pith.science/api/pith-number/MBCF2YU64WYWT6FHFLBWF2EAAQ/graph.json","events_json":"https://pith.science/api/pith-number/MBCF2YU64WYWT6FHFLBWF2EAAQ/events.json","paper":"https://pith.science/paper/MBCF2YU6"},"agent_actions":{"view_html":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ","download_json":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ.json","view_paper":"https://pith.science/paper/MBCF2YU6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.08044&json=true","fetch_graph":"https://pith.science/api/pith-number/MBCF2YU64WYWT6FHFLBWF2EAAQ/graph.json","fetch_events":"https://pith.science/api/pith-number/MBCF2YU64WYWT6FHFLBWF2EAAQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ/action/storage_attestation","attest_author":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ/action/author_attestation","sign_citation":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ/action/citation_signature","submit_replication":"https://pith.science/pith/MBCF2YU64WYWT6FHFLBWF2EAAQ/action/replication_record"}},"created_at":"2026-07-05T11:35:16.274505+00:00","updated_at":"2026-07-05T11:35:16.274505+00:00"}