{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TUAFI22KJSH5YIF65FVTYF42ZU","short_pith_number":"pith:TUAFI22K","schema_version":"1.0","canonical_sha256":"9d00546b4a4c8fdc20bee96b3c179acd1fff3e0074112cb02859c5ff6ac6feaf","source":{"kind":"arxiv","id":"2606.17296","version":1},"attestation_state":"computed","paper":{"title":"Pareto LoRA: Mitigating Modality Imbalance in Unified Multimodal Models via Pareto-Optimal Gradient Integration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Madhusudhanan Srinivasan, Mark Nutter, Radu Marculescu, Xiwen Wei","submitted_at":"2026-06-15T21:05:57Z","abstract_excerpt":"Unified multimodal models (UMMs) have recently emerged as a promising paradigm for integrating multimodal understanding and generation within a single autoregressive transformer. However, during multimodal instruction tuning, these models often exhibit pronounced modality imbalance: language gradients dominate optimization, thus leading to lower image generation quality, especially under parameter-efficient fine-tuning such as LoRA. In this work, we systematically analyze modality imbalance in LoRA-based fine-tuning of UMMs for interleaved text-image generation. We show that vision modality pe"},"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":"2606.17296","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-06-15T21:05:57Z","cross_cats_sorted":[],"title_canon_sha256":"f990c17b9e7c10e1188ffbe176d63c264122d0626ac4fbb81c4eb24ec0020086","abstract_canon_sha256":"629262e3561b4bf211a5dac6defb01eecc3a5a89cb1eefbec53c7255986e13bf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:10:08.398028Z","signature_b64":"4yjiKxp6EwLpCAB3t2sB2SRDQW36xxcnrUpgx0IB1kdP1rWFm02txCP8C+zmQTslwgMinlR+X0pmuiQ/GgXeDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9d00546b4a4c8fdc20bee96b3c179acd1fff3e0074112cb02859c5ff6ac6feaf","last_reissued_at":"2026-06-19T16:10:08.397585Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:10:08.397585Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pareto LoRA: Mitigating Modality Imbalance in Unified Multimodal Models via Pareto-Optimal Gradient Integration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Madhusudhanan Srinivasan, Mark Nutter, Radu Marculescu, Xiwen Wei","submitted_at":"2026-06-15T21:05:57Z","abstract_excerpt":"Unified multimodal models (UMMs) have recently emerged as a promising paradigm for integrating multimodal understanding and generation within a single autoregressive transformer. However, during multimodal instruction tuning, these models often exhibit pronounced modality imbalance: language gradients dominate optimization, thus leading to lower image generation quality, especially under parameter-efficient fine-tuning such as LoRA. In this work, we systematically analyze modality imbalance in LoRA-based fine-tuning of UMMs for interleaved text-image generation. We show that vision modality pe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.17296","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/2606.17296/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":"2606.17296","created_at":"2026-06-19T16:10:08.397651+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.17296v1","created_at":"2026-06-19T16:10:08.397651+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.17296","created_at":"2026-06-19T16:10:08.397651+00:00"},{"alias_kind":"pith_short_12","alias_value":"TUAFI22KJSH5","created_at":"2026-06-19T16:10:08.397651+00:00"},{"alias_kind":"pith_short_16","alias_value":"TUAFI22KJSH5YIF6","created_at":"2026-06-19T16:10:08.397651+00:00"},{"alias_kind":"pith_short_8","alias_value":"TUAFI22K","created_at":"2026-06-19T16:10:08.397651+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/TUAFI22KJSH5YIF65FVTYF42ZU","json":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU.json","graph_json":"https://pith.science/api/pith-number/TUAFI22KJSH5YIF65FVTYF42ZU/graph.json","events_json":"https://pith.science/api/pith-number/TUAFI22KJSH5YIF65FVTYF42ZU/events.json","paper":"https://pith.science/paper/TUAFI22K"},"agent_actions":{"view_html":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU","download_json":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU.json","view_paper":"https://pith.science/paper/TUAFI22K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.17296&json=true","fetch_graph":"https://pith.science/api/pith-number/TUAFI22KJSH5YIF65FVTYF42ZU/graph.json","fetch_events":"https://pith.science/api/pith-number/TUAFI22KJSH5YIF65FVTYF42ZU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU/action/storage_attestation","attest_author":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU/action/author_attestation","sign_citation":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU/action/citation_signature","submit_replication":"https://pith.science/pith/TUAFI22KJSH5YIF65FVTYF42ZU/action/replication_record"}},"created_at":"2026-06-19T16:10:08.397651+00:00","updated_at":"2026-06-19T16:10:08.397651+00:00"}