{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2JA5AUR3DWXDQBB7X7VL542WHI","short_pith_number":"pith:2JA5AUR3","schema_version":"1.0","canonical_sha256":"d241d0523b1dae38043fbfeabef3563a050244ea48df180cf0511a25754e6e62","source":{"kind":"arxiv","id":"2502.08482","version":1},"attestation_state":"computed","paper":{"title":"Enhancing Auto-regressive Chain-of-Thought through Loop-Aligned Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Di He, Jingjing Xu, Jun Zhang, Qifan Yu, Sijie Li, Xun Zhou, Zhenyu He","submitted_at":"2025-02-12T15:17:04Z","abstract_excerpt":"Chain-of-Thought (CoT) prompting has emerged as a powerful technique for enhancing language model's reasoning capabilities. However, generating long and correct CoT trajectories is challenging. Recent studies have demonstrated that Looped Transformers possess remarkable length generalization capabilities, but their limited generality and adaptability prevent them from serving as an alternative to auto-regressive solutions. To better leverage the strengths of Looped Transformers, we propose RELAY (REasoning through Loop Alignment iterativelY). Specifically, we align the steps of Chain-of-Though"},"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":"2502.08482","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-12T15:17:04Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"cd731038130daffa3ce329abb11da809cd9abcdb99c905adf8b0e58e0a585052","abstract_canon_sha256":"d06a72d5ac68ba78ce8923cb4cc0e66d9114d19e6a07d3a75c40caca61628651"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:13:23.697798Z","signature_b64":"F8OMpbM1eI6yAFgFhxKyrmhLGX88TWk6C2GYIi8yrXo3ddJUh1+E6+pz4BE+wwoNZPgo+1U30lrmtuO2vl/cBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d241d0523b1dae38043fbfeabef3563a050244ea48df180cf0511a25754e6e62","last_reissued_at":"2026-07-05T10:13:23.697206Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:13:23.697206Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Enhancing Auto-regressive Chain-of-Thought through Loop-Aligned Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Di He, Jingjing Xu, Jun Zhang, Qifan Yu, Sijie Li, Xun Zhou, Zhenyu He","submitted_at":"2025-02-12T15:17:04Z","abstract_excerpt":"Chain-of-Thought (CoT) prompting has emerged as a powerful technique for enhancing language model's reasoning capabilities. However, generating long and correct CoT trajectories is challenging. Recent studies have demonstrated that Looped Transformers possess remarkable length generalization capabilities, but their limited generality and adaptability prevent them from serving as an alternative to auto-regressive solutions. To better leverage the strengths of Looped Transformers, we propose RELAY (REasoning through Loop Alignment iterativelY). Specifically, we align the steps of Chain-of-Though"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.08482","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/2502.08482/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":"2502.08482","created_at":"2026-07-05T10:13:23.697282+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.08482v1","created_at":"2026-07-05T10:13:23.697282+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.08482","created_at":"2026-07-05T10:13:23.697282+00:00"},{"alias_kind":"pith_short_12","alias_value":"2JA5AUR3DWXD","created_at":"2026-07-05T10:13:23.697282+00:00"},{"alias_kind":"pith_short_16","alias_value":"2JA5AUR3DWXDQBB7","created_at":"2026-07-05T10:13:23.697282+00:00"},{"alias_kind":"pith_short_8","alias_value":"2JA5AUR3","created_at":"2026-07-05T10:13:23.697282+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18022","citing_title":"Recursive Scaling in Masked Diffusion Models","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01062","citing_title":"DAG-MoE: From Simple Mixture to Structural Aggregation in Mixture-of-Experts","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2508.16745","citing_title":"Beyond Memorization: Extending Reasoning Depth with Recurrence, Memory and Test-Time Compute Scaling","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2502.21074","citing_title":"CODI: Compressing Chain-of-Thought into Continuous Space via Self-Distillation","ref_index":133,"is_internal_anchor":false},{"citing_arxiv_id":"2511.08983","citing_title":"SpiralThinker: Latent Reasoning through an Iterative Process with Text-Latent Interleaving","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI","json":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI.json","graph_json":"https://pith.science/api/pith-number/2JA5AUR3DWXDQBB7X7VL542WHI/graph.json","events_json":"https://pith.science/api/pith-number/2JA5AUR3DWXDQBB7X7VL542WHI/events.json","paper":"https://pith.science/paper/2JA5AUR3"},"agent_actions":{"view_html":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI","download_json":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI.json","view_paper":"https://pith.science/paper/2JA5AUR3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.08482&json=true","fetch_graph":"https://pith.science/api/pith-number/2JA5AUR3DWXDQBB7X7VL542WHI/graph.json","fetch_events":"https://pith.science/api/pith-number/2JA5AUR3DWXDQBB7X7VL542WHI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI/action/storage_attestation","attest_author":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI/action/author_attestation","sign_citation":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI/action/citation_signature","submit_replication":"https://pith.science/pith/2JA5AUR3DWXDQBB7X7VL542WHI/action/replication_record"}},"created_at":"2026-07-05T10:13:23.697282+00:00","updated_at":"2026-07-05T10:13:23.697282+00:00"}