{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:7FQYK2WQTZRV6ZXG57SRYNM2K6","short_pith_number":"pith:7FQYK2WQ","schema_version":"1.0","canonical_sha256":"f961856ad09e635f66e6efe51c359a5796a88510818c594843bcf9c40f74653a","source":{"kind":"arxiv","id":"2202.12172","version":1},"attestation_state":"computed","paper":{"title":"Overcoming a Theoretical Limitation of Self-Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"David Chiang, Peter Cholak","submitted_at":"2022-02-24T16:14:29Z","abstract_excerpt":"Although transformers are remarkably effective for many tasks, there are some surprisingly easy-looking regular languages that they struggle with. Hahn shows that for languages where acceptance depends on a single input symbol, a transformer's classification decisions become less and less confident (that is, with cross-entropy approaching 1 bit per string) as input strings get longer and longer. We examine this limitation using two languages: PARITY, the language of bit strings with an odd number of 1s, and FIRST, the language of bit strings starting with a 1. We demonstrate three ways of over"},"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":"2202.12172","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-02-24T16:14:29Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"622b553e09a207bb802b02fa36d69216358feb51373c4bf76a362dc785276e8c","abstract_canon_sha256":"3481ce81942ba95b3e06f7b901e2e4c53efc08b0da2e566ac4e4bc535181854e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:59:50.711908Z","signature_b64":"Qx002yt0FWySQAiq47zmYRgIJt0LhUPlvZCQHAXlmSZTG87uDkyM6VutU6UJf3xJWGS8RsOHtPR/I7vZyfvSCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f961856ad09e635f66e6efe51c359a5796a88510818c594843bcf9c40f74653a","last_reissued_at":"2026-07-05T03:59:50.711480Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:59:50.711480Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Overcoming a Theoretical Limitation of Self-Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"David Chiang, Peter Cholak","submitted_at":"2022-02-24T16:14:29Z","abstract_excerpt":"Although transformers are remarkably effective for many tasks, there are some surprisingly easy-looking regular languages that they struggle with. Hahn shows that for languages where acceptance depends on a single input symbol, a transformer's classification decisions become less and less confident (that is, with cross-entropy approaching 1 bit per string) as input strings get longer and longer. We examine this limitation using two languages: PARITY, the language of bit strings with an odd number of 1s, and FIRST, the language of bit strings starting with a 1. We demonstrate three ways of over"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.12172","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/2202.12172/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":"2202.12172","created_at":"2026-07-05T03:59:50.711539+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.12172v1","created_at":"2026-07-05T03:59:50.711539+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.12172","created_at":"2026-07-05T03:59:50.711539+00:00"},{"alias_kind":"pith_short_12","alias_value":"7FQYK2WQTZRV","created_at":"2026-07-05T03:59:50.711539+00:00"},{"alias_kind":"pith_short_16","alias_value":"7FQYK2WQTZRV6ZXG","created_at":"2026-07-05T03:59:50.711539+00:00"},{"alias_kind":"pith_short_8","alias_value":"7FQYK2WQ","created_at":"2026-07-05T03:59:50.711539+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.08537","citing_title":"Meta-learning In-Context Enables Training-Free Cross Subject Brain Decoding","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6","json":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6.json","graph_json":"https://pith.science/api/pith-number/7FQYK2WQTZRV6ZXG57SRYNM2K6/graph.json","events_json":"https://pith.science/api/pith-number/7FQYK2WQTZRV6ZXG57SRYNM2K6/events.json","paper":"https://pith.science/paper/7FQYK2WQ"},"agent_actions":{"view_html":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6","download_json":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6.json","view_paper":"https://pith.science/paper/7FQYK2WQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.12172&json=true","fetch_graph":"https://pith.science/api/pith-number/7FQYK2WQTZRV6ZXG57SRYNM2K6/graph.json","fetch_events":"https://pith.science/api/pith-number/7FQYK2WQTZRV6ZXG57SRYNM2K6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6/action/storage_attestation","attest_author":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6/action/author_attestation","sign_citation":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6/action/citation_signature","submit_replication":"https://pith.science/pith/7FQYK2WQTZRV6ZXG57SRYNM2K6/action/replication_record"}},"created_at":"2026-07-05T03:59:50.711539+00:00","updated_at":"2026-07-05T03:59:50.711539+00:00"}