{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UUWHL3U55WVQ63OSMJ4T6JPEII","short_pith_number":"pith:UUWHL3U5","schema_version":"1.0","canonical_sha256":"a52c75ee9dedab0f6dd262793f25e442320dc12360d07c24b007f50fd4b518c5","source":{"kind":"arxiv","id":"2501.17745","version":2},"attestation_state":"computed","paper":{"title":"Dynamics of Transient Structure in In-Context Linear Regression Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Daniel Murfet, Jesse Hoogland, Liam Carroll, Matthew Farrugia-Roberts","submitted_at":"2025-01-29T16:32:14Z","abstract_excerpt":"Modern deep neural networks display striking examples of rich internal computational structure. Uncovering principles governing the development of such structure is a priority for the science of deep learning. In this paper, we explore the transient ridge phenomenon: when transformers are trained on in-context linear regression tasks with intermediate task diversity, they initially behave like ridge regression before specializing to the tasks in their training distribution. This transition from a general solution to a specialized solution is revealed by joint trajectory principal component ana"},"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":"2501.17745","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-01-29T16:32:14Z","cross_cats_sorted":[],"title_canon_sha256":"2e0a77493e39516923bf2626f4f8a8ab8c44324f4e2206a52c989fa7d4a96996","abstract_canon_sha256":"d0500e0f9325d13b77015d1d119bab6dd90d24d9d9d9c2f1df417c04115fb0d0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:07:48.556583Z","signature_b64":"4fwIdytVNoPz/d7GZuFTcLIMUg8/pXWwaHxHqkzCabK+9tnRs6/dXiHPRQUjIUIyQ/m0U4ZhSTKyRHzE/X95BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a52c75ee9dedab0f6dd262793f25e442320dc12360d07c24b007f50fd4b518c5","last_reissued_at":"2026-07-05T10:07:48.556094Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:07:48.556094Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dynamics of Transient Structure in In-Context Linear Regression Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Daniel Murfet, Jesse Hoogland, Liam Carroll, Matthew Farrugia-Roberts","submitted_at":"2025-01-29T16:32:14Z","abstract_excerpt":"Modern deep neural networks display striking examples of rich internal computational structure. Uncovering principles governing the development of such structure is a priority for the science of deep learning. In this paper, we explore the transient ridge phenomenon: when transformers are trained on in-context linear regression tasks with intermediate task diversity, they initially behave like ridge regression before specializing to the tasks in their training distribution. This transition from a general solution to a specialized solution is revealed by joint trajectory principal component ana"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.17745","kind":"arxiv","version":2},"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/2501.17745/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":"2501.17745","created_at":"2026-07-05T10:07:48.556149+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.17745v2","created_at":"2026-07-05T10:07:48.556149+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.17745","created_at":"2026-07-05T10:07:48.556149+00:00"},{"alias_kind":"pith_short_12","alias_value":"UUWHL3U55WVQ","created_at":"2026-07-05T10:07:48.556149+00:00"},{"alias_kind":"pith_short_16","alias_value":"UUWHL3U55WVQ63OS","created_at":"2026-07-05T10:07:48.556149+00:00"},{"alias_kind":"pith_short_8","alias_value":"UUWHL3U5","created_at":"2026-07-05T10:07:48.556149+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01302","citing_title":"Structure and Scale in Simplicial Sequence Modelling","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05076","citing_title":"High-Dimensional Statistics: Reflections on Progress and Open Problems","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18281","citing_title":"Temporal Task Diversity: Inductive Biases Under Non-Stationarity in Synthetic Sequence Modelling","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03780","citing_title":"Task Vector Geometry Underlies Dual Modes of Task Inference in Transformers","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05076","citing_title":"High-Dimensional Statistics: Reflections on Progress and Open Problems","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05115","citing_title":"Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior","ref_index":256,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18970","citing_title":"Mechanistic Anomaly Detection via Functional Attribution","ref_index":67,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII","json":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII.json","graph_json":"https://pith.science/api/pith-number/UUWHL3U55WVQ63OSMJ4T6JPEII/graph.json","events_json":"https://pith.science/api/pith-number/UUWHL3U55WVQ63OSMJ4T6JPEII/events.json","paper":"https://pith.science/paper/UUWHL3U5"},"agent_actions":{"view_html":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII","download_json":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII.json","view_paper":"https://pith.science/paper/UUWHL3U5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.17745&json=true","fetch_graph":"https://pith.science/api/pith-number/UUWHL3U55WVQ63OSMJ4T6JPEII/graph.json","fetch_events":"https://pith.science/api/pith-number/UUWHL3U55WVQ63OSMJ4T6JPEII/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII/action/storage_attestation","attest_author":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII/action/author_attestation","sign_citation":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII/action/citation_signature","submit_replication":"https://pith.science/pith/UUWHL3U55WVQ63OSMJ4T6JPEII/action/replication_record"}},"created_at":"2026-07-05T10:07:48.556149+00:00","updated_at":"2026-07-05T10:07:48.556149+00:00"}