{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:4FDCT5P4P5SLR4TMZAONGDXQRG","short_pith_number":"pith:4FDCT5P4","schema_version":"1.0","canonical_sha256":"e14629f5fc7f64b8f26cc81cd30ef0899d1f8f7398f9440832c3792465e71a92","source":{"kind":"arxiv","id":"2601.22495","version":2},"attestation_state":"computed","paper":{"title":"Gradual Fine-Tuning for Flow Matching Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Arindam Banerjee, Gudrun Thorkelsdottir","submitted_at":"2026-01-30T03:07:47Z","abstract_excerpt":"Fine-tuning flow matching models is a central challenge in settings with limited data, evolving distributions, or computational constraints. While recent work has produced significant advances, particularly in the area of reward-based fine-tuning, current methods fail to demonstrate both theoretical correctness as well as strong empirical results in terms of stability, efficiency, and diversity preservation. In this work, we propose Gradual Fine-Tuning (GFT), a simple yet principled annealing-based framework for fine-tuning flow generative models when only samples from the target distribution "},"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":"2601.22495","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-01-30T03:07:47Z","cross_cats_sorted":[],"title_canon_sha256":"ae4e3cfd15528f06c0a08ed1529dccdc4eae23bd42852ebc5c99c85ef46ad05b","abstract_canon_sha256":"2c371aca1ec87bcb83e230b997d94d337a590d6a1f70666f046a30ada64ebe93"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:09:54.227434Z","signature_b64":"qH0T93QHgwHo3H9jP6N3Gz/7OnbY5HKwvsditeD1QdXt1Uk38nUwN7OPKwmrW6TEQMyo4g++s/l8kj9Nm6XZAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e14629f5fc7f64b8f26cc81cd30ef0899d1f8f7398f9440832c3792465e71a92","last_reissued_at":"2026-06-19T16:09:54.227009Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:09:54.227009Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gradual Fine-Tuning for Flow Matching Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Arindam Banerjee, Gudrun Thorkelsdottir","submitted_at":"2026-01-30T03:07:47Z","abstract_excerpt":"Fine-tuning flow matching models is a central challenge in settings with limited data, evolving distributions, or computational constraints. While recent work has produced significant advances, particularly in the area of reward-based fine-tuning, current methods fail to demonstrate both theoretical correctness as well as strong empirical results in terms of stability, efficiency, and diversity preservation. In this work, we propose Gradual Fine-Tuning (GFT), a simple yet principled annealing-based framework for fine-tuning flow generative models when only samples from the target distribution "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.22495","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/2601.22495/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":"2601.22495","created_at":"2026-06-19T16:09:54.227064+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.22495v2","created_at":"2026-06-19T16:09:54.227064+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.22495","created_at":"2026-06-19T16:09:54.227064+00:00"},{"alias_kind":"pith_short_12","alias_value":"4FDCT5P4P5SL","created_at":"2026-06-19T16:09:54.227064+00:00"},{"alias_kind":"pith_short_16","alias_value":"4FDCT5P4P5SLR4TM","created_at":"2026-06-19T16:09:54.227064+00:00"},{"alias_kind":"pith_short_8","alias_value":"4FDCT5P4","created_at":"2026-06-19T16:09:54.227064+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/4FDCT5P4P5SLR4TMZAONGDXQRG","json":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG.json","graph_json":"https://pith.science/api/pith-number/4FDCT5P4P5SLR4TMZAONGDXQRG/graph.json","events_json":"https://pith.science/api/pith-number/4FDCT5P4P5SLR4TMZAONGDXQRG/events.json","paper":"https://pith.science/paper/4FDCT5P4"},"agent_actions":{"view_html":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG","download_json":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG.json","view_paper":"https://pith.science/paper/4FDCT5P4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.22495&json=true","fetch_graph":"https://pith.science/api/pith-number/4FDCT5P4P5SLR4TMZAONGDXQRG/graph.json","fetch_events":"https://pith.science/api/pith-number/4FDCT5P4P5SLR4TMZAONGDXQRG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG/action/storage_attestation","attest_author":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG/action/author_attestation","sign_citation":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG/action/citation_signature","submit_replication":"https://pith.science/pith/4FDCT5P4P5SLR4TMZAONGDXQRG/action/replication_record"}},"created_at":"2026-06-19T16:09:54.227064+00:00","updated_at":"2026-06-19T16:09:54.227064+00:00"}