{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:FGFNE7RQCHMLFHSZGDDYLPSBR4","short_pith_number":"pith:FGFNE7RQ","schema_version":"1.0","canonical_sha256":"298ad27e3011d8b29e5930c785be418f06e54132a0077de0970b3502d879a518","source":{"kind":"arxiv","id":"2312.03002","version":1},"attestation_state":"computed","paper":{"title":"The mechanistic basis of data dependence and abrupt learning in an in-context classification task","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Gautam Reddy","submitted_at":"2023-12-03T20:53:41Z","abstract_excerpt":"Transformer models exhibit in-context learning: the ability to accurately predict the response to a novel query based on illustrative examples in the input sequence. In-context learning contrasts with traditional in-weights learning of query-output relationships. What aspects of the training data distribution and architecture favor in-context vs in-weights learning? Recent work has shown that specific distributional properties inherent in language, such as burstiness, large dictionaries and skewed rank-frequency distributions, control the trade-off or simultaneous appearance of these two forms"},"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":"2312.03002","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2023-12-03T20:53:41Z","cross_cats_sorted":[],"title_canon_sha256":"8418f8435cc48e985fa868cf39e2b703d09a2c0f639c2cf4eba5c7d5f468b2ee","abstract_canon_sha256":"586c03b149a8c77a77da693a96f43bacc4a5734287f5f7872f3bddf0a1f23985"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:20:56.706895Z","signature_b64":"uJWuYJUfITKSH1N+BYrqQcoUAIhzMHpeog1CDHn2MtWcWx1CIuxC3SV2NBFYDTcQ32Bu3xdjEooaUdckY8HlAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"298ad27e3011d8b29e5930c785be418f06e54132a0077de0970b3502d879a518","last_reissued_at":"2026-07-05T07:20:56.706440Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:20:56.706440Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The mechanistic basis of data dependence and abrupt learning in an in-context classification task","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Gautam Reddy","submitted_at":"2023-12-03T20:53:41Z","abstract_excerpt":"Transformer models exhibit in-context learning: the ability to accurately predict the response to a novel query based on illustrative examples in the input sequence. In-context learning contrasts with traditional in-weights learning of query-output relationships. What aspects of the training data distribution and architecture favor in-context vs in-weights learning? Recent work has shown that specific distributional properties inherent in language, such as burstiness, large dictionaries and skewed rank-frequency distributions, control the trade-off or simultaneous appearance of these two forms"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.03002","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/2312.03002/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":"2312.03002","created_at":"2026-07-05T07:20:56.706500+00:00"},{"alias_kind":"arxiv_version","alias_value":"2312.03002v1","created_at":"2026-07-05T07:20:56.706500+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.03002","created_at":"2026-07-05T07:20:56.706500+00:00"},{"alias_kind":"pith_short_12","alias_value":"FGFNE7RQCHML","created_at":"2026-07-05T07:20:56.706500+00:00"},{"alias_kind":"pith_short_16","alias_value":"FGFNE7RQCHMLFHSZ","created_at":"2026-07-05T07:20:56.706500+00:00"},{"alias_kind":"pith_short_8","alias_value":"FGFNE7RQ","created_at":"2026-07-05T07:20:56.706500+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06621","citing_title":"Fingerprint, Not Blueprint: How Positional Schemes Set the Default Spectral Algebra of Attention","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2606.26021","citing_title":"Privacy Vulnerabilities of Attention Layers in Tabular Foundation Models and Protection of High-Risk Queries","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20299","citing_title":"Mechanisms of Misgeneralization in Physical Sequence Modeling","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2506.04289","citing_title":"Relational reasoning and inductive bias in transformers and large language models","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21691","citing_title":"There Will Be a Scientific Theory of Deep Learning","ref_index":163,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21632","citing_title":"To See the Unseen: on the Generalization Ability of Transformers in Symbolic Reasoning","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4","json":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4.json","graph_json":"https://pith.science/api/pith-number/FGFNE7RQCHMLFHSZGDDYLPSBR4/graph.json","events_json":"https://pith.science/api/pith-number/FGFNE7RQCHMLFHSZGDDYLPSBR4/events.json","paper":"https://pith.science/paper/FGFNE7RQ"},"agent_actions":{"view_html":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4","download_json":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4.json","view_paper":"https://pith.science/paper/FGFNE7RQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2312.03002&json=true","fetch_graph":"https://pith.science/api/pith-number/FGFNE7RQCHMLFHSZGDDYLPSBR4/graph.json","fetch_events":"https://pith.science/api/pith-number/FGFNE7RQCHMLFHSZGDDYLPSBR4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4/action/storage_attestation","attest_author":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4/action/author_attestation","sign_citation":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4/action/citation_signature","submit_replication":"https://pith.science/pith/FGFNE7RQCHMLFHSZGDDYLPSBR4/action/replication_record"}},"created_at":"2026-07-05T07:20:56.706500+00:00","updated_at":"2026-07-05T07:20:56.706500+00:00"}