{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:KENTDCMHNLV2WWTMX5ZLQNGTKZ","short_pith_number":"pith:KENTDCMH","schema_version":"1.0","canonical_sha256":"511b3189876aebab5a6cbf72b834d3566788ed9b4f564a2dfb2e2ef690118e9f","source":{"kind":"arxiv","id":"2210.11610","version":2},"attestation_state":"computed","paper":{"title":"Large Language Models Can Self-Improve","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hongkun Yu, Jiawei Han, Jiaxin Huang, Le Hou, Shixiang Shane Gu, Xuezhi Wang, Yuexin Wu","submitted_at":"2022-10-20T21:53:54Z","abstract_excerpt":"Large Language Models (LLMs) have achieved excellent performances in various tasks. However, fine-tuning an LLM requires extensive supervision. Human, on the other hand, may improve their reasoning abilities by self-thinking without external inputs. In this work, we demonstrate that an LLM is also capable of self-improving with only unlabeled datasets. We use a pre-trained LLM to generate \"high-confidence\" rationale-augmented answers for unlabeled questions using Chain-of-Thought prompting and self-consistency, and fine-tune the LLM using those self-generated solutions as target outputs. We sh"},"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":"2210.11610","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-10-20T21:53:54Z","cross_cats_sorted":[],"title_canon_sha256":"768340164ea52e8134df65e7104c55cacdb444963b4dfe006c7d3e1585fd3c84","abstract_canon_sha256":"277f369b4d507b788ab4e3b566c713a4f7185465c78a3b78e70fa740a835a527"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T16:54:03.321312Z","signature_b64":"R1p52OxZNWQZvV7Un7itsiq+IAvZrD/C6HmxO9faRqk1hV7uLGIZ5s6OYogobRCjMKVO9swY1QhrnjjL6ebQAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"511b3189876aebab5a6cbf72b834d3566788ed9b4f564a2dfb2e2ef690118e9f","last_reissued_at":"2026-05-18T16:54:03.318232Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T16:54:03.318232Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large Language Models Can Self-Improve","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hongkun Yu, Jiawei Han, Jiaxin Huang, Le Hou, Shixiang Shane Gu, Xuezhi Wang, Yuexin Wu","submitted_at":"2022-10-20T21:53:54Z","abstract_excerpt":"Large Language Models (LLMs) have achieved excellent performances in various tasks. However, fine-tuning an LLM requires extensive supervision. Human, on the other hand, may improve their reasoning abilities by self-thinking without external inputs. In this work, we demonstrate that an LLM is also capable of self-improving with only unlabeled datasets. We use a pre-trained LLM to generate \"high-confidence\" rationale-augmented answers for unlabeled questions using Chain-of-Thought prompting and self-consistency, and fine-tune the LLM using those self-generated solutions as target outputs. We sh"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.11610","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":""},"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":"2210.11610","created_at":"2026-05-18T16:54:03.318375+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.11610v2","created_at":"2026-05-18T16:54:03.318375+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.11610","created_at":"2026-05-18T16:54:03.318375+00:00"},{"alias_kind":"pith_short_12","alias_value":"KENTDCMHNLV2","created_at":"2026-05-18T16:54:03.318375+00:00"},{"alias_kind":"pith_short_16","alias_value":"KENTDCMHNLV2WWTM","created_at":"2026-05-18T16:54:03.318375+00:00"},{"alias_kind":"pith_short_8","alias_value":"KENTDCMH","created_at":"2026-05-18T16:54:03.318375+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":24,"internal_anchor_count":24,"sample":[{"citing_arxiv_id":"2302.12039","citing_title":"Natural Language Processing in the Legal Domain","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2402.01411","citing_title":"CodePori: Large-Scale System for Autonomous Software Development Using Multi-Agent Technology","ref_index":42,"is_internal_anchor":true},{"citing_arxiv_id":"2305.02301","citing_title":"Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes","ref_index":80,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17187","citing_title":"PluRule: A Benchmark for Moderating Pluralistic Communities on Social Media","ref_index":91,"is_internal_anchor":true},{"citing_arxiv_id":"2309.17452","citing_title":"ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2509.11295","citing_title":"The Prompt Engineering Report Distilled: Quick Start Guide for Life Sciences","ref_index":61,"is_internal_anchor":true},{"citing_arxiv_id":"2509.11295","citing_title":"The Prompt Engineering Report Distilled: Quick Start Guide for Life Sciences","ref_index":62,"is_internal_anchor":true},{"citing_arxiv_id":"2505.15134","citing_title":"The Unreasonable Effectiveness of Entropy Minimization in LLM Reasoning","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2408.00724","citing_title":"Inference Scaling Laws: An Empirical Analysis of Compute-Optimal Inference for Problem-Solving with Language Models","ref_index":157,"is_internal_anchor":true},{"citing_arxiv_id":"2409.12917","citing_title":"Training Language Models to Self-Correct via Reinforcement Learning","ref_index":129,"is_internal_anchor":true},{"citing_arxiv_id":"2401.05459","citing_title":"Personal LLM Agents: Insights and Survey about the Capability, Efficiency and Security","ref_index":243,"is_internal_anchor":true},{"citing_arxiv_id":"2303.09014","citing_title":"ART: Automatic multi-step reasoning and tool-use for large language models","ref_index":135,"is_internal_anchor":true},{"citing_arxiv_id":"2411.10440","citing_title":"LLaVA-CoT: Let Vision Language Models Reason Step-by-Step","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2309.16797","citing_title":"Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution","ref_index":154,"is_internal_anchor":true},{"citing_arxiv_id":"2309.00267","citing_title":"RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback","ref_index":122,"is_internal_anchor":true},{"citing_arxiv_id":"2406.06592","citing_title":"Improve Mathematical Reasoning in Language Models by Automated Process Supervision","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2309.12284","citing_title":"MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Models","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2604.24625","citing_title":"Meta-CoT: Enhancing Granularity and Generalization in Image Editing","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2310.03714","citing_title":"DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelines","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2604.21611","citing_title":"Process Supervision via Verbal Critique Improves Reasoning in Large Language Models","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2604.19015","citing_title":"FedProxy: Federated Fine-Tuning of LLMs via Proxy SLMs and Heterogeneity-Aware Fusion","ref_index":94,"is_internal_anchor":true},{"citing_arxiv_id":"2604.09791","citing_title":"Pioneer Agent: Continual Improvement of Small Language Models in Production","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2507.20534","citing_title":"Kimi K2: Open Agentic Intelligence","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2604.17433","citing_title":"Self-Consistency from Only Two Samples: CoT-PoT Ensembling for Efficient LLM Reasoning","ref_index":54,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ","json":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ.json","graph_json":"https://pith.science/api/pith-number/KENTDCMHNLV2WWTMX5ZLQNGTKZ/graph.json","events_json":"https://pith.science/api/pith-number/KENTDCMHNLV2WWTMX5ZLQNGTKZ/events.json","paper":"https://pith.science/paper/KENTDCMH"},"agent_actions":{"view_html":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ","download_json":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ.json","view_paper":"https://pith.science/paper/KENTDCMH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.11610&json=true","fetch_graph":"https://pith.science/api/pith-number/KENTDCMHNLV2WWTMX5ZLQNGTKZ/graph.json","fetch_events":"https://pith.science/api/pith-number/KENTDCMHNLV2WWTMX5ZLQNGTKZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ/action/storage_attestation","attest_author":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ/action/author_attestation","sign_citation":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ/action/citation_signature","submit_replication":"https://pith.science/pith/KENTDCMHNLV2WWTMX5ZLQNGTKZ/action/replication_record"}},"created_at":"2026-05-18T16:54:03.318375+00:00","updated_at":"2026-05-18T16:54:03.318375+00:00"}