{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:HJOEX2J2W6X67EFAMF5NFS53RY","short_pith_number":"pith:HJOEX2J2","schema_version":"1.0","canonical_sha256":"3a5c4be93ab7afef90a0617ad2cbbb8e1b32f21cd82cf2e69343d3b318a3b569","source":{"kind":"arxiv","id":"2310.00035","version":2},"attestation_state":"computed","paper":{"title":"LoRA ensembles for large language model fine-tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Laurence Aitchison, Maja Rudolph, Xi Wang","submitted_at":"2023-09-29T16:38:38Z","abstract_excerpt":"Finetuned LLMs often exhibit poor uncertainty quantification, manifesting as overconfidence, poor calibration, and unreliable prediction results on test data or out-of-distribution samples. One approach commonly used in vision for alleviating this issue is a deep ensemble, which constructs an ensemble by training the same model multiple times using different random initializations. However, there is a huge challenge to ensembling LLMs: the most effective LLMs are very, very large. Keeping a single LLM in memory is already challenging enough: keeping an ensemble of e.g. 5 LLMs in memory is impo"},"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":"2310.00035","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-09-29T16:38:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"3b6f578e3aa4b9d0e7d2bbe2cc7c21bedd3347cad81f0e2fdea43488f9c6b60b","abstract_canon_sha256":"6817a7cf3282400dd6a02991a506b8647c59a69de3bcd6987e3e220298b4a87e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:57:26.970442Z","signature_b64":"bZoLeIZIjlEsyVRh8GwcFfPUr6AMZ/WMIQat3jQeoqAsrTdkjozcQ6m57k5EPiKIiVzuZvioZRoCUNQsrii2Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3a5c4be93ab7afef90a0617ad2cbbb8e1b32f21cd82cf2e69343d3b318a3b569","last_reissued_at":"2026-07-05T06:57:26.969970Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:57:26.969970Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LoRA ensembles for large language model fine-tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Laurence Aitchison, Maja Rudolph, Xi Wang","submitted_at":"2023-09-29T16:38:38Z","abstract_excerpt":"Finetuned LLMs often exhibit poor uncertainty quantification, manifesting as overconfidence, poor calibration, and unreliable prediction results on test data or out-of-distribution samples. One approach commonly used in vision for alleviating this issue is a deep ensemble, which constructs an ensemble by training the same model multiple times using different random initializations. However, there is a huge challenge to ensembling LLMs: the most effective LLMs are very, very large. Keeping a single LLM in memory is already challenging enough: keeping an ensemble of e.g. 5 LLMs in memory is impo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.00035","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/2310.00035/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":"2310.00035","created_at":"2026-07-05T06:57:26.970037+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.00035v2","created_at":"2026-07-05T06:57:26.970037+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.00035","created_at":"2026-07-05T06:57:26.970037+00:00"},{"alias_kind":"pith_short_12","alias_value":"HJOEX2J2W6X6","created_at":"2026-07-05T06:57:26.970037+00:00"},{"alias_kind":"pith_short_16","alias_value":"HJOEX2J2W6X67EFA","created_at":"2026-07-05T06:57:26.970037+00:00"},{"alias_kind":"pith_short_8","alias_value":"HJOEX2J2","created_at":"2026-07-05T06:57:26.970037+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22792","citing_title":"The Origins of Stochasticity: Comprehensive Investigations on Uncertainty Quantification for Large Language Models","ref_index":106,"is_internal_anchor":false},{"citing_arxiv_id":"2607.02182","citing_title":"Bayesian Sparse Low-Rank Adaptation for Large Language Model Uncertainty Estimation","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29184","citing_title":"BaRA: Bayesian Adaptive Rank Allocation for Parameter-Efficient Fine-Tuning","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28920","citing_title":"Conf-Gen: Conformal Uncertainty Quantification for Generative Models","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2410.06431","citing_title":"Functional-level Uncertainty Quantification for Calibrated Fine-tuning on LLMs","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20408","citing_title":"Spectral Souping: A Unified Framework for Online Preference Alignment","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2506.10060","citing_title":"Textual Bayes: Quantifying Prompt Uncertainty in LLM-Based Systems","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2604.03388","citing_title":"Scalable Variational Bayesian Fine-Tuning of LLMs via Orthogonalized Low-Rank Adapters","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11328","citing_title":"Epistemic Uncertainty for Test-Time Discovery","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08110","citing_title":"BaLoRA: Bayesian Low-Rank Adaptation of Large Scale Models","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06330","citing_title":"Fine-Tuning Small Language Models for Solution-Oriented Windows Event Log Analysis","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04738","citing_title":"Fine-Tuning Integrity for Modern Neural Networks: Structured Drift Proofs via Norm, Rank, and Sparsity Certificates","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17112","citing_title":"Complementing Self-Consistency with Cross-Model Disagreement for Uncertainty Quantification","ref_index":44,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY","json":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY.json","graph_json":"https://pith.science/api/pith-number/HJOEX2J2W6X67EFAMF5NFS53RY/graph.json","events_json":"https://pith.science/api/pith-number/HJOEX2J2W6X67EFAMF5NFS53RY/events.json","paper":"https://pith.science/paper/HJOEX2J2"},"agent_actions":{"view_html":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY","download_json":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY.json","view_paper":"https://pith.science/paper/HJOEX2J2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.00035&json=true","fetch_graph":"https://pith.science/api/pith-number/HJOEX2J2W6X67EFAMF5NFS53RY/graph.json","fetch_events":"https://pith.science/api/pith-number/HJOEX2J2W6X67EFAMF5NFS53RY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY/action/storage_attestation","attest_author":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY/action/author_attestation","sign_citation":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY/action/citation_signature","submit_replication":"https://pith.science/pith/HJOEX2J2W6X67EFAMF5NFS53RY/action/replication_record"}},"created_at":"2026-07-05T06:57:26.970037+00:00","updated_at":"2026-07-05T06:57:26.970037+00:00"}