{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JHN7TO6W4NTILBKAXGEYPAAC5V","short_pith_number":"pith:JHN7TO6W","schema_version":"1.0","canonical_sha256":"49dbf9bbd6e366858540b989878002ed679a526ba795d44cd2a274a184e8434e","source":{"kind":"arxiv","id":"2504.04383","version":2},"attestation_state":"computed","paper":{"title":"Retro-Search: Exploring Untaken Paths for Deeper and Efficient Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Bryan Catanzaro, David Acuna, Hyunwoo Kim, Jaehun Jung, Mohammad Shoeybi, Mostofa Patwary, Niklas Muennighoff, Seungju Han, Shrimai Prabhumoye, Ximing Lu, Yejin Choi","submitted_at":"2025-04-06T06:23:27Z","abstract_excerpt":"Large reasoning models exhibit remarkable reasoning capabilities via long, elaborate reasoning trajectories. Supervised fine-tuning on such reasoning traces, also known as distillation, can be a cost-effective way to boost reasoning capabilities of student models. However, empirical observations reveal that these reasoning trajectories are often suboptimal, switching excessively between different lines of thought, resulting in under-thinking, over-thinking, and even degenerate responses. We introduce Retro-Search, an MCTS-inspired search algorithm, for distilling higher quality reasoning paths"},"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":"2504.04383","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-04-06T06:23:27Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"edd7198c5f4490875c41dfc9e243b0e3a7935693e2d793057261b875c060104c","abstract_canon_sha256":"a832fcce66980737dc422c56de988efd920fbd092addfd0338b6eb5f11ac90c6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:49:16.605057Z","signature_b64":"uqNJT3UjRvV+xd2H6XYqL0SKkPNTwVmuPNh5YE8VAL0v6aggv4l10twpH6CkojsI1EwWz8e6qDnX5/ZhIEAnBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"49dbf9bbd6e366858540b989878002ed679a526ba795d44cd2a274a184e8434e","last_reissued_at":"2026-07-05T10:49:16.604529Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:49:16.604529Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Retro-Search: Exploring Untaken Paths for Deeper and Efficient Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Bryan Catanzaro, David Acuna, Hyunwoo Kim, Jaehun Jung, Mohammad Shoeybi, Mostofa Patwary, Niklas Muennighoff, Seungju Han, Shrimai Prabhumoye, Ximing Lu, Yejin Choi","submitted_at":"2025-04-06T06:23:27Z","abstract_excerpt":"Large reasoning models exhibit remarkable reasoning capabilities via long, elaborate reasoning trajectories. Supervised fine-tuning on such reasoning traces, also known as distillation, can be a cost-effective way to boost reasoning capabilities of student models. However, empirical observations reveal that these reasoning trajectories are often suboptimal, switching excessively between different lines of thought, resulting in under-thinking, over-thinking, and even degenerate responses. We introduce Retro-Search, an MCTS-inspired search algorithm, for distilling higher quality reasoning paths"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.04383","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/2504.04383/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":"2504.04383","created_at":"2026-07-05T10:49:16.604596+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.04383v2","created_at":"2026-07-05T10:49:16.604596+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.04383","created_at":"2026-07-05T10:49:16.604596+00:00"},{"alias_kind":"pith_short_12","alias_value":"JHN7TO6W4NTI","created_at":"2026-07-05T10:49:16.604596+00:00"},{"alias_kind":"pith_short_16","alias_value":"JHN7TO6W4NTILBKA","created_at":"2026-07-05T10:49:16.604596+00:00"},{"alias_kind":"pith_short_8","alias_value":"JHN7TO6W","created_at":"2026-07-05T10:49:16.604596+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18089","citing_title":"From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning","ref_index":114,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00206","citing_title":"Quantized Reasoning Models Think They Need to Think Longer, but They Do Not","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01111","citing_title":"When Less is Enough: Efficient Inference via Collaborative Reasoning","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17304","citing_title":"Efficient Test-Time Scaling via Temporal Reasoning Aggregation","ref_index":63,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V","json":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V.json","graph_json":"https://pith.science/api/pith-number/JHN7TO6W4NTILBKAXGEYPAAC5V/graph.json","events_json":"https://pith.science/api/pith-number/JHN7TO6W4NTILBKAXGEYPAAC5V/events.json","paper":"https://pith.science/paper/JHN7TO6W"},"agent_actions":{"view_html":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V","download_json":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V.json","view_paper":"https://pith.science/paper/JHN7TO6W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.04383&json=true","fetch_graph":"https://pith.science/api/pith-number/JHN7TO6W4NTILBKAXGEYPAAC5V/graph.json","fetch_events":"https://pith.science/api/pith-number/JHN7TO6W4NTILBKAXGEYPAAC5V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V/action/storage_attestation","attest_author":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V/action/author_attestation","sign_citation":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V/action/citation_signature","submit_replication":"https://pith.science/pith/JHN7TO6W4NTILBKAXGEYPAAC5V/action/replication_record"}},"created_at":"2026-07-05T10:49:16.604596+00:00","updated_at":"2026-07-05T10:49:16.604596+00:00"}