{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4STNHU57EO4FF7TGZ4ME4LA5J7","short_pith_number":"pith:4STNHU57","schema_version":"1.0","canonical_sha256":"e4a6d3d3bf23b852fe66cf184e2c1d4fecc36e1409806e89ec73076c4f063624","source":{"kind":"arxiv","id":"2505.13775","version":4},"attestation_state":"computed","paper":{"title":"Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Atharva Gundawar, Karthik Valmeekam, Kaya Stechly, Subbarao Kambhampati, Vardhan Palod","submitted_at":"2025-05-19T23:29:23Z","abstract_excerpt":"Recent impressive results from large reasoning models have been interpreted as a triumph of Chain of Thought (CoT), and especially of the process of training on CoTs sampled from base LLMs in order to help find new reasoning patterns. While these traces certainly seem to help model performance, it is not clear how they influence it, with some works ascribing semantics to them and others cautioning against relying on them as transparent and faithful proxies of the model's internal computational process. To systematically investigate the role of end-user semantics of derivational traces, we set "},"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":"2505.13775","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-05-19T23:29:23Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"861d2a1eb534e74bc740a672ddb46d6b33aa371b8b3b6449b748ccd8f15a76e8","abstract_canon_sha256":"1a823d59aae7fc2a1b53f44c9b6953b95efc241b531564fcac0e475c807ef29c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:04:48.416975Z","signature_b64":"eE9kOj9krMMK7+emB9XNlb7+uXnMSx8t5CQQDapcBVhcEiurWqwFPjyM5311SziH4PaqP6PMV5VPPluaVsryCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e4a6d3d3bf23b852fe66cf184e2c1d4fecc36e1409806e89ec73076c4f063624","last_reissued_at":"2026-05-27T01:04:48.416299Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:04:48.416299Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokens","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Atharva Gundawar, Karthik Valmeekam, Kaya Stechly, Subbarao Kambhampati, Vardhan Palod","submitted_at":"2025-05-19T23:29:23Z","abstract_excerpt":"Recent impressive results from large reasoning models have been interpreted as a triumph of Chain of Thought (CoT), and especially of the process of training on CoTs sampled from base LLMs in order to help find new reasoning patterns. While these traces certainly seem to help model performance, it is not clear how they influence it, with some works ascribing semantics to them and others cautioning against relying on them as transparent and faithful proxies of the model's internal computational process. To systematically investigate the role of end-user semantics of derivational traces, we set "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.13775","kind":"arxiv","version":4},"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/2505.13775/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":"2505.13775","created_at":"2026-05-27T01:04:48.416394+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.13775v4","created_at":"2026-05-27T01:04:48.416394+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.13775","created_at":"2026-05-27T01:04:48.416394+00:00"},{"alias_kind":"pith_short_12","alias_value":"4STNHU57EO4F","created_at":"2026-05-27T01:04:48.416394+00:00"},{"alias_kind":"pith_short_16","alias_value":"4STNHU57EO4FF7TG","created_at":"2026-05-27T01:04:48.416394+00:00"},{"alias_kind":"pith_short_8","alias_value":"4STNHU57","created_at":"2026-05-27T01:04:48.416394+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":10,"sample":[{"citing_arxiv_id":"2605.10930","citing_title":"Evaluating the False Trust Engendered by LLM Explanations","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17565","citing_title":"Generalization or Memorization? Brittleness Testing for Chess-Trained Language Models","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2508.01191","citing_title":"Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2510.25426","citing_title":"Implicature in Interaction: Understanding Implicature Improves Alignment in Human-LLM Interaction","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11206","citing_title":"Instructions Shape Production of Language, not Processing","ref_index":213,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11206","citing_title":"Instructions Shape Production of Language, not Processing","ref_index":213,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10930","citing_title":"Evaluating the False Trust Engendered by LLM Explanations","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09519","citing_title":"Weighted Rules under the Stable Model Semantics","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07776","citing_title":"Tracing Uncertainty in Language Model \"Reasoning\"","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2604.17957","citing_title":"Process Reward Models Meet Planning: Generating Precise and Scalable Datasets for Step-Level Rewards","ref_index":49,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7","json":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7.json","graph_json":"https://pith.science/api/pith-number/4STNHU57EO4FF7TGZ4ME4LA5J7/graph.json","events_json":"https://pith.science/api/pith-number/4STNHU57EO4FF7TGZ4ME4LA5J7/events.json","paper":"https://pith.science/paper/4STNHU57"},"agent_actions":{"view_html":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7","download_json":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7.json","view_paper":"https://pith.science/paper/4STNHU57","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.13775&json=true","fetch_graph":"https://pith.science/api/pith-number/4STNHU57EO4FF7TGZ4ME4LA5J7/graph.json","fetch_events":"https://pith.science/api/pith-number/4STNHU57EO4FF7TGZ4ME4LA5J7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7/action/storage_attestation","attest_author":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7/action/author_attestation","sign_citation":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7/action/citation_signature","submit_replication":"https://pith.science/pith/4STNHU57EO4FF7TGZ4ME4LA5J7/action/replication_record"}},"created_at":"2026-05-27T01:04:48.416394+00:00","updated_at":"2026-05-27T01:04:48.416394+00:00"}