{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SJQK5XTMDT46ZNHLD4ZWQO4MWK","short_pith_number":"pith:SJQK5XTM","schema_version":"1.0","canonical_sha256":"9260aede6c1cf9ecb4eb1f33683b8cb285fd2ce1f2c8b85496fe8360fbbe263b","source":{"kind":"arxiv","id":"2606.08051","version":1},"attestation_state":"computed","paper":{"title":"How Small Can You Go? LoRA Fine-Tuning 270M-8B Models for Merchant Information Extraction in Financial Transactions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Benjamin Barrett, Donghao Huang, Tomas Drietomsky, Zhaoxia Wang","submitted_at":"2026-06-06T08:32:38Z","abstract_excerpt":"Financial transaction processing requires extracting structured merchant information from noisy, abbreviated bank transaction strings at scale. Our current production system, a LoRA-fine-tuned LLaMA 3.1-8B, achieves 96.95% F1 on this task, but deploying 8-billion-parameter models imposes prohibitive memory, latency, and cost constraints. To identify more efficient alternatives, we conduct a deployment-focused study of 24 model variants spanning four model families: Gemma 3 (270M, 1B, 4B), Qwen 3.5 (0.8B, 2B, 4B), Aya (3.35B), and LLaMA 3.1-8B, systematically evaluating accuracy, inference thro"},"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":"2606.08051","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-06T08:32:38Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e00191e3c4f466aaec34eca5dc18b36e2e551de64874bf2fe953380e92021a94","abstract_canon_sha256":"7b8ac3e8b244d090c2192f105529d9475a0c346ad06b9702c7761cd7b8d6c38d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:24.563734Z","signature_b64":"jJvb3eX2X4LWoXMCrw59JfZJyKuLiClxwMrgs66tCxtXIPyFgWcSNfBXERc1f6/9vRGaQz021Vq0N6XsP+/7BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9260aede6c1cf9ecb4eb1f33683b8cb285fd2ce1f2c8b85496fe8360fbbe263b","last_reissued_at":"2026-06-09T01:05:24.563054Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:24.563054Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How Small Can You Go? LoRA Fine-Tuning 270M-8B Models for Merchant Information Extraction in Financial Transactions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Benjamin Barrett, Donghao Huang, Tomas Drietomsky, Zhaoxia Wang","submitted_at":"2026-06-06T08:32:38Z","abstract_excerpt":"Financial transaction processing requires extracting structured merchant information from noisy, abbreviated bank transaction strings at scale. Our current production system, a LoRA-fine-tuned LLaMA 3.1-8B, achieves 96.95% F1 on this task, but deploying 8-billion-parameter models imposes prohibitive memory, latency, and cost constraints. To identify more efficient alternatives, we conduct a deployment-focused study of 24 model variants spanning four model families: Gemma 3 (270M, 1B, 4B), Qwen 3.5 (0.8B, 2B, 4B), Aya (3.35B), and LLaMA 3.1-8B, systematically evaluating accuracy, inference thro"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08051","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/2606.08051/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":"2606.08051","created_at":"2026-06-09T01:05:24.563150+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.08051v1","created_at":"2026-06-09T01:05:24.563150+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08051","created_at":"2026-06-09T01:05:24.563150+00:00"},{"alias_kind":"pith_short_12","alias_value":"SJQK5XTMDT46","created_at":"2026-06-09T01:05:24.563150+00:00"},{"alias_kind":"pith_short_16","alias_value":"SJQK5XTMDT46ZNHL","created_at":"2026-06-09T01:05:24.563150+00:00"},{"alias_kind":"pith_short_8","alias_value":"SJQK5XTM","created_at":"2026-06-09T01:05:24.563150+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/SJQK5XTMDT46ZNHLD4ZWQO4MWK","json":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK.json","graph_json":"https://pith.science/api/pith-number/SJQK5XTMDT46ZNHLD4ZWQO4MWK/graph.json","events_json":"https://pith.science/api/pith-number/SJQK5XTMDT46ZNHLD4ZWQO4MWK/events.json","paper":"https://pith.science/paper/SJQK5XTM"},"agent_actions":{"view_html":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK","download_json":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK.json","view_paper":"https://pith.science/paper/SJQK5XTM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.08051&json=true","fetch_graph":"https://pith.science/api/pith-number/SJQK5XTMDT46ZNHLD4ZWQO4MWK/graph.json","fetch_events":"https://pith.science/api/pith-number/SJQK5XTMDT46ZNHLD4ZWQO4MWK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK/action/storage_attestation","attest_author":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK/action/author_attestation","sign_citation":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK/action/citation_signature","submit_replication":"https://pith.science/pith/SJQK5XTMDT46ZNHLD4ZWQO4MWK/action/replication_record"}},"created_at":"2026-06-09T01:05:24.563150+00:00","updated_at":"2026-06-09T01:05:24.563150+00:00"}