{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TFQPEI6JL5RSM2535Z3WU6EF43","short_pith_number":"pith:TFQPEI6J","schema_version":"1.0","canonical_sha256":"9960f223c95f63266bbbee776a7885e6f41068c1b2bba2780e0fbe32b785f418","source":{"kind":"arxiv","id":"2508.03999","version":1},"attestation_state":"computed","paper":{"title":"Tensorized Clustered LoRA Merging for Multi-Task Interference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bingbing Wen, Fengran Mo, Guojun Liang, Jian-Yun Nie, Jinghan Zhang, Prayag Tiwari, Zhan Su","submitted_at":"2025-08-06T01:26:43Z","abstract_excerpt":"Despite the success of the monolithic dense paradigm of large language models (LLMs), the LoRA adapters offer an efficient solution by fine-tuning small task-specific modules and merging them with the base model. However, in multi-task settings, merging LoRA adapters trained on heterogeneous sources frequently causes \\textit{task interference}, degrading downstream performance. To address this, we propose a tensorized clustered LoRA (TC-LoRA) library targeting to address the task interference at the \\textit{text-level} and \\textit{parameter-level}. At the \\textit{text-level}, we cluster the tr"},"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":"2508.03999","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-06T01:26:43Z","cross_cats_sorted":[],"title_canon_sha256":"fbcff81bfe49983e46e39b647f72a727d415a4926dcbbb41e90c3e55b37e78c5","abstract_canon_sha256":"021c46de9d53080eafa5bb06ea3331a11da82aaef6b8086bc4b88eea867281f1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:49:18.751006Z","signature_b64":"CJQkttvmrBQyirvDZms82eAZg8dqvjKKU2/i3kTooQ/+xXtxxa7Szcvp27iVX0W0R7b2MOXD2zFTuYQfnB1QAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9960f223c95f63266bbbee776a7885e6f41068c1b2bba2780e0fbe32b785f418","last_reissued_at":"2026-07-05T11:49:18.750578Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:49:18.750578Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tensorized Clustered LoRA Merging for Multi-Task Interference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bingbing Wen, Fengran Mo, Guojun Liang, Jian-Yun Nie, Jinghan Zhang, Prayag Tiwari, Zhan Su","submitted_at":"2025-08-06T01:26:43Z","abstract_excerpt":"Despite the success of the monolithic dense paradigm of large language models (LLMs), the LoRA adapters offer an efficient solution by fine-tuning small task-specific modules and merging them with the base model. However, in multi-task settings, merging LoRA adapters trained on heterogeneous sources frequently causes \\textit{task interference}, degrading downstream performance. To address this, we propose a tensorized clustered LoRA (TC-LoRA) library targeting to address the task interference at the \\textit{text-level} and \\textit{parameter-level}. At the \\textit{text-level}, we cluster the tr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.03999","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/2508.03999/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":"2508.03999","created_at":"2026-07-05T11:49:18.750637+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.03999v1","created_at":"2026-07-05T11:49:18.750637+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.03999","created_at":"2026-07-05T11:49:18.750637+00:00"},{"alias_kind":"pith_short_12","alias_value":"TFQPEI6JL5RS","created_at":"2026-07-05T11:49:18.750637+00:00"},{"alias_kind":"pith_short_16","alias_value":"TFQPEI6JL5RSM253","created_at":"2026-07-05T11:49:18.750637+00:00"},{"alias_kind":"pith_short_8","alias_value":"TFQPEI6J","created_at":"2026-07-05T11:49:18.750637+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/TFQPEI6JL5RSM2535Z3WU6EF43","json":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43.json","graph_json":"https://pith.science/api/pith-number/TFQPEI6JL5RSM2535Z3WU6EF43/graph.json","events_json":"https://pith.science/api/pith-number/TFQPEI6JL5RSM2535Z3WU6EF43/events.json","paper":"https://pith.science/paper/TFQPEI6J"},"agent_actions":{"view_html":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43","download_json":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43.json","view_paper":"https://pith.science/paper/TFQPEI6J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.03999&json=true","fetch_graph":"https://pith.science/api/pith-number/TFQPEI6JL5RSM2535Z3WU6EF43/graph.json","fetch_events":"https://pith.science/api/pith-number/TFQPEI6JL5RSM2535Z3WU6EF43/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43/action/storage_attestation","attest_author":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43/action/author_attestation","sign_citation":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43/action/citation_signature","submit_replication":"https://pith.science/pith/TFQPEI6JL5RSM2535Z3WU6EF43/action/replication_record"}},"created_at":"2026-07-05T11:49:18.750637+00:00","updated_at":"2026-07-05T11:49:18.750637+00:00"}