{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UQ525LCX7QGGCIZDQ64QISOLW4","short_pith_number":"pith:UQ525LCX","schema_version":"1.0","canonical_sha256":"a43baeac57fc0c61232387b90449cbb70e6612338536bdedd98106295699f2f9","source":{"kind":"arxiv","id":"2606.02907","version":1},"attestation_state":"computed","paper":{"title":"Linear Probes Detect Task Format, Not Reasoning Mode in Language Model Hidden States","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Aman Chadha, Divya Chaudhary, Subramanyam Sahoo, Vinija Jain","submitted_at":"2026-06-01T21:22:15Z","abstract_excerpt":"Linear probing of large language model (LLM) hidden states is widely used to claim that models learn distinct representations for different reasoning types. We test this by probing Qwen3-14B on three benchmarks spanning the classical trichotomy: LogiQA 2.0 (deductive), ARC-Challenge (inductive), and $\\alpha$NLI (abductive). At layer 32 of 40, linear probes achieve 100\\% cross-validated accuracy with well-separated geometry (intrinsic dimensionalities: 20.6, 28.5, 33.6; convex hull contamination $\\leq$1.5\\%). However, this separation is entirely driven by format confounds. Residualizing source "},"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":"2606.02907","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-01T21:22:15Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"07b56a3ac8e1ffa1ce425410651d836e6e8bf6d56dbca99ab263d490fa1231c3","abstract_canon_sha256":"ce6059e379b691510a28cd871857641b6606d69f7cd3da387c4a10b034e50731"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T01:05:26.236954Z","signature_b64":"nE0MUPl4t9fjzhhkH2wxED2UheDlkgsg3yuG9M0diyMfEbc/W/fs+XaZsEcYbZCBGI3nUP8wfnbfLOYpzDaCBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a43baeac57fc0c61232387b90449cbb70e6612338536bdedd98106295699f2f9","last_reissued_at":"2026-06-03T01:05:26.236541Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T01:05:26.236541Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Linear Probes Detect Task Format, Not Reasoning Mode in Language Model Hidden States","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Aman Chadha, Divya Chaudhary, Subramanyam Sahoo, Vinija Jain","submitted_at":"2026-06-01T21:22:15Z","abstract_excerpt":"Linear probing of large language model (LLM) hidden states is widely used to claim that models learn distinct representations for different reasoning types. We test this by probing Qwen3-14B on three benchmarks spanning the classical trichotomy: LogiQA 2.0 (deductive), ARC-Challenge (inductive), and $\\alpha$NLI (abductive). At layer 32 of 40, linear probes achieve 100\\% cross-validated accuracy with well-separated geometry (intrinsic dimensionalities: 20.6, 28.5, 33.6; convex hull contamination $\\leq$1.5\\%). However, this separation is entirely driven by format confounds. Residualizing source "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.02907","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/2606.02907/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":"2606.02907","created_at":"2026-06-03T01:05:26.236603+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.02907v1","created_at":"2026-06-03T01:05:26.236603+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.02907","created_at":"2026-06-03T01:05:26.236603+00:00"},{"alias_kind":"pith_short_12","alias_value":"UQ525LCX7QGG","created_at":"2026-06-03T01:05:26.236603+00:00"},{"alias_kind":"pith_short_16","alias_value":"UQ525LCX7QGGCIZD","created_at":"2026-06-03T01:05:26.236603+00:00"},{"alias_kind":"pith_short_8","alias_value":"UQ525LCX","created_at":"2026-06-03T01:05:26.236603+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4","json":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4.json","graph_json":"https://pith.science/api/pith-number/UQ525LCX7QGGCIZDQ64QISOLW4/graph.json","events_json":"https://pith.science/api/pith-number/UQ525LCX7QGGCIZDQ64QISOLW4/events.json","paper":"https://pith.science/paper/UQ525LCX"},"agent_actions":{"view_html":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4","download_json":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4.json","view_paper":"https://pith.science/paper/UQ525LCX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.02907&json=true","fetch_graph":"https://pith.science/api/pith-number/UQ525LCX7QGGCIZDQ64QISOLW4/graph.json","fetch_events":"https://pith.science/api/pith-number/UQ525LCX7QGGCIZDQ64QISOLW4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4/action/storage_attestation","attest_author":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4/action/author_attestation","sign_citation":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4/action/citation_signature","submit_replication":"https://pith.science/pith/UQ525LCX7QGGCIZDQ64QISOLW4/action/replication_record"}},"created_at":"2026-06-03T01:05:26.236603+00:00","updated_at":"2026-06-03T01:05:26.236603+00:00"}