{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NSSWMX5QWAOOX2YELH4QR5UMQ2","short_pith_number":"pith:NSSWMX5Q","schema_version":"1.0","canonical_sha256":"6ca5665fb0b01cebeb0459f908f68c86aa19a7b1a77f48d8a7a2effceff5fe0b","source":{"kind":"arxiv","id":"2605.22870","version":1},"attestation_state":"computed","paper":{"title":"The Readout Shortcut: Positional Number Copying Dominates Arithmetic CoT Readout in Small Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ming Liu","submitted_at":"2026-05-20T00:32:41Z","abstract_excerpt":"Chain-of-thought (CoT) prompting is necessary for arithmetic in small language models, yet shuffling its steps preserves most performance. What does CoT contribute if not logical sequencing? In three 1-3B instruction-tuned LMs on GSM8K, we isolate the answer-readout stage via prefix completion and identify a positional shortcut: the model copies whichever number occupies the trailing position before the answer delimiter, regardless of intermediate reasoning. Gold-answer presence accounts for 54-92 pp of accuracy (89-92% of each model's teacher-forcing ceiling); even on incorrect items, the fin"},"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":"2605.22870","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-20T00:32:41Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"097018dc9a1e825c48ec1f05d8f85775b3147d9b5d5264792205ea9e19276a84","abstract_canon_sha256":"d8759a0e1be7ec33543298f1953823d13e73e332575a105bdfb6b33344438808"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:01:27.888854Z","signature_b64":"K0Y+yoi9XGoRtSB1jYYrP2ONj0Eatvc51o0LsWYpuQ6Gh3BpkJuwgg+xH/g76/sO1iFvkiugc8iow74X1raBDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6ca5665fb0b01cebeb0459f908f68c86aa19a7b1a77f48d8a7a2effceff5fe0b","last_reissued_at":"2026-05-25T02:01:27.888027Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:01:27.888027Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Readout Shortcut: Positional Number Copying Dominates Arithmetic CoT Readout in Small Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ming Liu","submitted_at":"2026-05-20T00:32:41Z","abstract_excerpt":"Chain-of-thought (CoT) prompting is necessary for arithmetic in small language models, yet shuffling its steps preserves most performance. What does CoT contribute if not logical sequencing? In three 1-3B instruction-tuned LMs on GSM8K, we isolate the answer-readout stage via prefix completion and identify a positional shortcut: the model copies whichever number occupies the trailing position before the answer delimiter, regardless of intermediate reasoning. Gold-answer presence accounts for 54-92 pp of accuracy (89-92% of each model's teacher-forcing ceiling); even on incorrect items, the fin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.22870","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/2605.22870/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":"2605.22870","created_at":"2026-05-25T02:01:27.888173+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.22870v1","created_at":"2026-05-25T02:01:27.888173+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.22870","created_at":"2026-05-25T02:01:27.888173+00:00"},{"alias_kind":"pith_short_12","alias_value":"NSSWMX5QWAOO","created_at":"2026-05-25T02:01:27.888173+00:00"},{"alias_kind":"pith_short_16","alias_value":"NSSWMX5QWAOOX2YE","created_at":"2026-05-25T02:01:27.888173+00:00"},{"alias_kind":"pith_short_8","alias_value":"NSSWMX5Q","created_at":"2026-05-25T02:01:27.888173+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/NSSWMX5QWAOOX2YELH4QR5UMQ2","json":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2.json","graph_json":"https://pith.science/api/pith-number/NSSWMX5QWAOOX2YELH4QR5UMQ2/graph.json","events_json":"https://pith.science/api/pith-number/NSSWMX5QWAOOX2YELH4QR5UMQ2/events.json","paper":"https://pith.science/paper/NSSWMX5Q"},"agent_actions":{"view_html":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2","download_json":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2.json","view_paper":"https://pith.science/paper/NSSWMX5Q","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.22870&json=true","fetch_graph":"https://pith.science/api/pith-number/NSSWMX5QWAOOX2YELH4QR5UMQ2/graph.json","fetch_events":"https://pith.science/api/pith-number/NSSWMX5QWAOOX2YELH4QR5UMQ2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2/action/storage_attestation","attest_author":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2/action/author_attestation","sign_citation":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2/action/citation_signature","submit_replication":"https://pith.science/pith/NSSWMX5QWAOOX2YELH4QR5UMQ2/action/replication_record"}},"created_at":"2026-05-25T02:01:27.888173+00:00","updated_at":"2026-05-25T02:01:27.888173+00:00"}