{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AZW3A26SGOOLVXWNTQLQCWXB5I","short_pith_number":"pith:AZW3A26S","schema_version":"1.0","canonical_sha256":"066db06bd2339cbadecd9c17015ae1ea2818e15075f058debe5b67bdff0effb4","source":{"kind":"arxiv","id":"2503.10167","version":2},"attestation_state":"computed","paper":{"title":"\"Well, Keep Thinking\": Enhancing LLM Reasoning with Adaptive Injection Decoding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hyunbin Jin, Je Won Yeom, Seunghyun Bae, Taesup Kim","submitted_at":"2025-03-13T08:46:32Z","abstract_excerpt":"Large language models (LLMs) exhibit strong reasoning abilities, often attributed to few-shot or zero-shot chain-of-thought (CoT) prompting. While effective, these methods require labor-intensive prompt engineering, raising the question of whether reasoning can be induced without reliance on explicit prompts. In this work, we unlock the reasoning capabilities of LLMs without explicit prompting. Inspired by zero-shot CoT and CoT-decoding, we propose a novel decoding strategy that systematically nudges LLMs to continue reasoning, thereby preventing immature reasoning processes. Specifically, we "},"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":"2503.10167","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-13T08:46:32Z","cross_cats_sorted":[],"title_canon_sha256":"2d412a4f67c6543db5752ddd5587d5a6c02597f4bbf5d247da98c7c04bad9f79","abstract_canon_sha256":"3223eba9f098f5553429ace49b72c8277cc0f7e6e97bf51ab93a0790b82f966d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:33:09.595137Z","signature_b64":"R0r0a9VxVvRO+JAn6s5vtGnYP60kembQentzZKn62cza/60W+TJe5+tWODMWlW2kclWFl88aNJmN+3Ofe5CCDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"066db06bd2339cbadecd9c17015ae1ea2818e15075f058debe5b67bdff0effb4","last_reissued_at":"2026-07-05T10:33:09.594091Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:33:09.594091Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"\"Well, Keep Thinking\": Enhancing LLM Reasoning with Adaptive Injection Decoding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hyunbin Jin, Je Won Yeom, Seunghyun Bae, Taesup Kim","submitted_at":"2025-03-13T08:46:32Z","abstract_excerpt":"Large language models (LLMs) exhibit strong reasoning abilities, often attributed to few-shot or zero-shot chain-of-thought (CoT) prompting. While effective, these methods require labor-intensive prompt engineering, raising the question of whether reasoning can be induced without reliance on explicit prompts. In this work, we unlock the reasoning capabilities of LLMs without explicit prompting. Inspired by zero-shot CoT and CoT-decoding, we propose a novel decoding strategy that systematically nudges LLMs to continue reasoning, thereby preventing immature reasoning processes. Specifically, we "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.10167","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/2503.10167/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":"2503.10167","created_at":"2026-07-05T10:33:09.594227+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.10167v2","created_at":"2026-07-05T10:33:09.594227+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.10167","created_at":"2026-07-05T10:33:09.594227+00:00"},{"alias_kind":"pith_short_12","alias_value":"AZW3A26SGOOL","created_at":"2026-07-05T10:33:09.594227+00:00"},{"alias_kind":"pith_short_16","alias_value":"AZW3A26SGOOLVXWN","created_at":"2026-07-05T10:33:09.594227+00:00"},{"alias_kind":"pith_short_8","alias_value":"AZW3A26S","created_at":"2026-07-05T10:33:09.594227+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11470","citing_title":"The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes","ref_index":106,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15725","citing_title":"Reasoning-targeted Jailbreak Attacks on Large Reasoning Models via Semantic Triggers and Psychological Framing","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I","json":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I.json","graph_json":"https://pith.science/api/pith-number/AZW3A26SGOOLVXWNTQLQCWXB5I/graph.json","events_json":"https://pith.science/api/pith-number/AZW3A26SGOOLVXWNTQLQCWXB5I/events.json","paper":"https://pith.science/paper/AZW3A26S"},"agent_actions":{"view_html":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I","download_json":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I.json","view_paper":"https://pith.science/paper/AZW3A26S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.10167&json=true","fetch_graph":"https://pith.science/api/pith-number/AZW3A26SGOOLVXWNTQLQCWXB5I/graph.json","fetch_events":"https://pith.science/api/pith-number/AZW3A26SGOOLVXWNTQLQCWXB5I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I/action/storage_attestation","attest_author":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I/action/author_attestation","sign_citation":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I/action/citation_signature","submit_replication":"https://pith.science/pith/AZW3A26SGOOLVXWNTQLQCWXB5I/action/replication_record"}},"created_at":"2026-07-05T10:33:09.594227+00:00","updated_at":"2026-07-05T10:33:09.594227+00:00"}