{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ROJBQUG7WJ3ADQQJGC7JG7IE6D","short_pith_number":"pith:ROJBQUG7","schema_version":"1.0","canonical_sha256":"8b921850dfb27601c20930be937d04f0e036ec9b5777e12d7af602dc5ef84661","source":{"kind":"arxiv","id":"2512.20845","version":2},"attestation_state":"computed","paper":{"title":"MAR:Multi-Agent Reflexion Improves Reasoning Abilities in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.MA"],"primary_cat":"cs.AI","authors_text":"Daniel Dosti, Grace Wu, Honghao Zhang, Onat Ozer, Vivi De La Rue, Yuchen Wang","submitted_at":"2025-12-23T23:47:31Z","abstract_excerpt":"LLMs have shown the capacity to improve their performance on reasoning tasks through reflecting on their mistakes, and acting with these reflections in mind. However, continual reflections of the same LLM onto itself exhibit degeneration of thought, where the LLM continues to repeat the same errors again and again even with the knowledge that its wrong. To address this problem, we instead introduce multi-agent with multi-persona debators as the method to generate reflections. Through out extensive experimentation, we've found that the leads to better diversity of in the reflections generated b"},"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":"2512.20845","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-12-23T23:47:31Z","cross_cats_sorted":["cs.MA"],"title_canon_sha256":"d421dd05b8f58ec3d65abc39e2942d74cd3561f0c7998bec4af357523063062f","abstract_canon_sha256":"97e8e9fcde58325a05e46ba03d061df46766d4337997a299eba9e906cda6953e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:46.748593Z","signature_b64":"J7D31G6r5QCUaxfJAN1oa9WAu874sDCQc7y/8iUuTaN1kk64gXLGt31dqz7s6nrImBK03iEocNP2IRCs/ZoUDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8b921850dfb27601c20930be937d04f0e036ec9b5777e12d7af602dc5ef84661","last_reissued_at":"2026-06-09T01:05:46.748167Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:46.748167Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MAR:Multi-Agent Reflexion Improves Reasoning Abilities in LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.MA"],"primary_cat":"cs.AI","authors_text":"Daniel Dosti, Grace Wu, Honghao Zhang, Onat Ozer, Vivi De La Rue, Yuchen Wang","submitted_at":"2025-12-23T23:47:31Z","abstract_excerpt":"LLMs have shown the capacity to improve their performance on reasoning tasks through reflecting on their mistakes, and acting with these reflections in mind. However, continual reflections of the same LLM onto itself exhibit degeneration of thought, where the LLM continues to repeat the same errors again and again even with the knowledge that its wrong. To address this problem, we instead introduce multi-agent with multi-persona debators as the method to generate reflections. Through out extensive experimentation, we've found that the leads to better diversity of in the reflections generated b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.20845","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/2512.20845/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":"2512.20845","created_at":"2026-06-09T01:05:46.748232+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.20845v2","created_at":"2026-06-09T01:05:46.748232+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.20845","created_at":"2026-06-09T01:05:46.748232+00:00"},{"alias_kind":"pith_short_12","alias_value":"ROJBQUG7WJ3A","created_at":"2026-06-09T01:05:46.748232+00:00"},{"alias_kind":"pith_short_16","alias_value":"ROJBQUG7WJ3ADQQJ","created_at":"2026-06-09T01:05:46.748232+00:00"},{"alias_kind":"pith_short_8","alias_value":"ROJBQUG7","created_at":"2026-06-09T01:05:46.748232+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2603.09002","citing_title":"Security Considerations for Multi-agent Systems","ref_index":145,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09121","citing_title":"A Communication-Theoretic Framework for LLM Agents: Cost-Aware Adaptive Reliability","ref_index":57,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06734","citing_title":"TEC: A Collection of Human Trial-and-error Trajectories for Problem Solving","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D","json":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D.json","graph_json":"https://pith.science/api/pith-number/ROJBQUG7WJ3ADQQJGC7JG7IE6D/graph.json","events_json":"https://pith.science/api/pith-number/ROJBQUG7WJ3ADQQJGC7JG7IE6D/events.json","paper":"https://pith.science/paper/ROJBQUG7"},"agent_actions":{"view_html":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D","download_json":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D.json","view_paper":"https://pith.science/paper/ROJBQUG7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.20845&json=true","fetch_graph":"https://pith.science/api/pith-number/ROJBQUG7WJ3ADQQJGC7JG7IE6D/graph.json","fetch_events":"https://pith.science/api/pith-number/ROJBQUG7WJ3ADQQJGC7JG7IE6D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D/action/storage_attestation","attest_author":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D/action/author_attestation","sign_citation":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D/action/citation_signature","submit_replication":"https://pith.science/pith/ROJBQUG7WJ3ADQQJGC7JG7IE6D/action/replication_record"}},"created_at":"2026-06-09T01:05:46.748232+00:00","updated_at":"2026-06-09T01:05:46.748232+00:00"}