{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UPFS2BOU2RIUDU6A3XFN5CEZC7","short_pith_number":"pith:UPFS2BOU","schema_version":"1.0","canonical_sha256":"a3cb2d05d4d45141d3c0ddcade889917ef9e197830ea67dff2539ef92b75f36c","source":{"kind":"arxiv","id":"2502.12215","version":2},"attestation_state":"computed","paper":{"title":"Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Qinyuan Cheng, Xipeng Qiu, Yunhua Zhou, Zhangyue Yin, Zhiyuan Zeng","submitted_at":"2025-02-17T07:21:11Z","abstract_excerpt":"The advent of test-time scaling in large language models (LLMs), exemplified by OpenAI's o1 series, has advanced reasoning capabilities by scaling computational resource allocation during inference. While successors like QwQ, Deepseek-R1 (R1) and LIMO replicate these advancements, whether these models truly possess test-time scaling capabilities remains underexplored. This study found that longer CoTs of these o1-like models do not consistently enhance accuracy; in fact, correct solutions are often shorter than incorrect ones for the same questions. Further investigation shows this phenomenon "},"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":"2502.12215","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-17T07:21:11Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"e54fb2362fa491f23f9ccd34853e006e4248c17005cf8d4724756489c1e81cdc","abstract_canon_sha256":"a152c3cf11ce1c045bc911fec3f36e74a1387bd90c8840018be6a66329fc5fd0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:22:57.051421Z","signature_b64":"ORCgC20KdUqFZ1uq+THd8vcZPTii+qyGICJt8k4q8P5QxVCEZmOh93sWJm0TuLUtOi0JNQSI7O2/xbzMvtvGBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a3cb2d05d4d45141d3c0ddcade889917ef9e197830ea67dff2539ef92b75f36c","last_reissued_at":"2026-07-05T10:22:57.050740Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:22:57.050740Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Revisiting the Test-Time Scaling of o1-like Models: Do they Truly Possess Test-Time Scaling Capabilities?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Qinyuan Cheng, Xipeng Qiu, Yunhua Zhou, Zhangyue Yin, Zhiyuan Zeng","submitted_at":"2025-02-17T07:21:11Z","abstract_excerpt":"The advent of test-time scaling in large language models (LLMs), exemplified by OpenAI's o1 series, has advanced reasoning capabilities by scaling computational resource allocation during inference. While successors like QwQ, Deepseek-R1 (R1) and LIMO replicate these advancements, whether these models truly possess test-time scaling capabilities remains underexplored. This study found that longer CoTs of these o1-like models do not consistently enhance accuracy; in fact, correct solutions are often shorter than incorrect ones for the same questions. Further investigation shows this phenomenon "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.12215","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/2502.12215/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":"2502.12215","created_at":"2026-07-05T10:22:57.050823+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.12215v2","created_at":"2026-07-05T10:22:57.050823+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.12215","created_at":"2026-07-05T10:22:57.050823+00:00"},{"alias_kind":"pith_short_12","alias_value":"UPFS2BOU2RIU","created_at":"2026-07-05T10:22:57.050823+00:00"},{"alias_kind":"pith_short_16","alias_value":"UPFS2BOU2RIUDU6A","created_at":"2026-07-05T10:22:57.050823+00:00"},{"alias_kind":"pith_short_8","alias_value":"UPFS2BOU","created_at":"2026-07-05T10:22:57.050823+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2508.13654","citing_title":"Input-Time Scaling: Adding Noise and Irrelevance into Less-Is-More Drastically Improves Reasoning Performance and Efficiency","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2508.10164","citing_title":"Pruning Long Chain-of-Thought of Large Reasoning Models via Small-Scale Preference Optimization","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10172","citing_title":"V-ABS: Action-Observer Driven Beam Search for Dynamic Visual Reasoning","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01194","citing_title":"VLA-ATTC: Adaptive Test-Time Compute for VLA Models with Relative Action Critic Model","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10449","citing_title":"AdverMCTS: Combating Pseudo-Correctness in Code Generation via Adversarial Monte Carlo Tree Search","ref_index":57,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7","json":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7.json","graph_json":"https://pith.science/api/pith-number/UPFS2BOU2RIUDU6A3XFN5CEZC7/graph.json","events_json":"https://pith.science/api/pith-number/UPFS2BOU2RIUDU6A3XFN5CEZC7/events.json","paper":"https://pith.science/paper/UPFS2BOU"},"agent_actions":{"view_html":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7","download_json":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7.json","view_paper":"https://pith.science/paper/UPFS2BOU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.12215&json=true","fetch_graph":"https://pith.science/api/pith-number/UPFS2BOU2RIUDU6A3XFN5CEZC7/graph.json","fetch_events":"https://pith.science/api/pith-number/UPFS2BOU2RIUDU6A3XFN5CEZC7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7/action/storage_attestation","attest_author":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7/action/author_attestation","sign_citation":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7/action/citation_signature","submit_replication":"https://pith.science/pith/UPFS2BOU2RIUDU6A3XFN5CEZC7/action/replication_record"}},"created_at":"2026-07-05T10:22:57.050823+00:00","updated_at":"2026-07-05T10:22:57.050823+00:00"}