{"as_of":"2026-08-19T14:34:00Z","caps":{"database_statements":6,"inbound":100,"outbound":100},"context_digest":"sha256:df6aec8548a6c4b2b0b51c5f94020228f7b7e2be67f7cfe731ba20bc157826a6","coverage":[{"denominator":50,"lane":"reference_resolution","note":"Typed states for the displayed outbound observations.","records_observed":50,"source":"paper_references, paper_reference_links","source_observed_at":"2026-07-13T06:19:30.027337Z","state":"measured"},{"denominator":50,"lane":"standing_notices","note":"One-hop event checks from named stored sources.","records_observed":50,"source":"scholarly_work_events, retraction_status_cache","source_observed_at":"2026-08-19T06:32:44.657259+00:00","state":"measured"},{"denominator":0,"lane":"inbound_itemization","note":"Pith citing papers itemized under the disclosed page cap.","records_observed":0,"source":"paper_references, paper_reference_links","source_observed_at":null,"state":"measured"},{"denominator":1,"lane":"external_citation_measurements","note":"A source-named dated measurement, never combined with another source.","records_observed":0,"source":"cited_works","source_observed_at":null,"state":"measured"}],"external_citation_measurements":[],"inbound":[],"links":{"evidence":"/evidence","html":"/paper/2607.08843/citation-record","integrity":"/paper/2607.08843/integrity","json":"/paper/2607.08843/citation-record.json","paper":"/paper/2607.08843"},"outbound":[{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Linguistic Regularities in Continuous Space Word Representations","venue":null,"work_id":null,"year":2013},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":1,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:f84eaaeda2d5088753ca398f74b9e655c90f4bdcff45550b6f49ffe9972abd88","observation_id":"30c6bb84-d694-451c-bfe9-addd39f84650","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2309.00941","last_updated":"2023-09-07T20:36:48Z","snapshot_observed_at":"2026-08-16T15:03:48.794640Z","submitted_at":"2023-09-02T13:37:34Z","title":"Emergent Linear Representations in World Models of Self-Supervised Sequence Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2309.00941","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Emergent Linear Representations in World Models of Self-Supervised Sequence Models, September 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":2,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2309.00941","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:b379d49eb080fd81bfba8cf30a6811607770431af3bee377d4f5c8eb68332460","observation_id":"fd4f8eb0-375e-4397-aab1-6c786969f6ab","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2311.03658","last_updated":"2024-07-17T22:24:27Z","snapshot_observed_at":"2026-08-16T23:25:38.049210Z","submitted_at":"2023-11-07T01:59:11Z","title":"The Linear Representation Hypothesis and the Geometry of Large Language Models","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2311.03658","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"The Linear Representation Hypothesis and the Geometry of Large Language Models, July 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":3,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2311.03658","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:d449e62660a91edcd2a8290d0a8057d9a04b72e5d7771c144600f37d249a76ca","observation_id":"252022a9-0203-40a8-aa44-4da326c6249a","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"On the Origins of Linear Representations in Large Language Models","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":4,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:8aefc68fbc788c3329c2ac91ec5e2da53a5ee66e85139fb4af27ab3039bf185b","observation_id":"fac94f7f-e3a6-44e7-913f-73aaaacc5b06","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2406.01506","last_updated":"2025-02-18T02:23:45Z","snapshot_observed_at":"2026-08-17T00:18:50.876112Z","submitted_at":"2024-06-03T16:34:01Z","title":"The Geometry of Categorical and Hierarchical Concepts in Large Language Models","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2406.01506","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"The Geometry of Categorical and Hier- archical Concepts in Large Language Models, February 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":5,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2406.01506","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:e1d637b07365585178a70f3112cdd4d7c0903d88e49574538a38a00d5cb65250","observation_id":"fe0fa536-efc1-47cb-93b3-95c739fa632e","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Benna, Mattia Rigotti, Jérôme Munuera, Stefano Fusi, and C","venue":null,"work_id":null,"year":2020},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":6,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:95037cecab7e8fa34ebef0b0760475b273042275644fca9010abcde8174089f7","observation_id":"22ac3703-ed4e-4c5b-9896-eda2f0d148da","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Rodgers, Randy M","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":7,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:7d4275d47d58a7c7d1d8caa503fa205d9060df9f31698829937944be7b40937b","observation_id":"be575075-af3c-4313-8b73-0a3e85523cc0","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Shin, Wenbo Tang, and Shantanu P","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":8,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:b9ec220cc24ad63dd1a0f77033712b89efb8a35589c4b516ba3dc18dfcd47179","observation_id":"a42811b6-ba57-42e9-8544-d81eb6d34bb0","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Neural representational geometries reflect behavioral differences in monkeys and recurrent neural networks.Nature Communications, 15(1):6479, August 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":9,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:1d884e62b62cc0ac7f262a83c4036f34dea7b2867bafa9f7f508f2daebff4f80","observation_id":"6fc8b8dd-e75c-48c0-a890-a65845aa9b7d","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Courellis, Juri Minxha, Araceli R","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":10,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:e03bde88f9ff2b3bdc7bbc08e0f9084c66f10d3e51b860b0fe1beafdf1daf2b8","observation_id":"e282ee65-6909-4cd3-8545-7396ea5e823d","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Boyle, Lorenzo Posani, Sarah Irfan, Steven A","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":11,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:8c11841c3cd3650b3c1caeb89840d9d053f9ed4b664b005800cf8298e3f0afe9","observation_id":"cf602ec4-8883-46a8-91b7-a5a3f705427f","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":12,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:7e2fd509b1d16510c02fd16979ab7b8b908d8b1c496fb3d8e26da0dffd1dceb1","observation_id":"64dbde20-d3e1-48c7-b431-af0a4caa3c3d","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Schoonover, Andrew J","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":13,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:e1a7c26e6011738c9623e881c4fc7076b1318a6762d3c6570eff29f2377ffe61","observation_id":"eaa97cbb-fa31-43a2-8766-d6b1caacc3a0","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2305.01610","last_updated":"2023-06-02T21:52:17Z","snapshot_observed_at":"2026-08-17T05:06:29.353999Z","submitted_at":"2023-05-02T17:13:55Z","title":"Finding Neurons in a Haystack: Case Studies with Sparse Probing","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2305.01610","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Finding Neurons in a Haystack: Case Studies with Sparse Probing, June 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":14,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2305.01610","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:da7b242a35ee65cd9cece40030c1421221c96741140934f602e96481c5250d9c","observation_id":"a618b0e9-6ab8-47c6-96d7-5ce7574e489d","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"De- tecting Strategic Deception with Linear Probes","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":15,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:ee9cb95c730ad9d2c5929bb0aa20063d0a14ce02ebf67e380cfc1293d8d2f44a","observation_id":"d961e035-f4eb-415e-aa6a-4a1e2ff84e5f","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Vazquez, Ulisse Mini, and Monte MacDiarmid","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":16,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:b46df3cf7777678ca148a7b3440e091ca5d1ac8f94407488afb9f9365a56c549","observation_id":"fab6a33c-284f-4540-8774-98da733686c6","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2308.10248","last_updated":"2024-10-10T13:20:13Z","snapshot_observed_at":"2026-07-06T02:11:23.670680Z","submitted_at":"2023-08-20T12:21:05Z","title":"Steering Language Models With Activation Engineering","version":5},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2308.10248","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":17,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2308.10248","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:ef7a3cbfc34f4d7b6abe5fb38844f0d66c39192a9d81b0f02c794e1babc96105","observation_id":"0a2f29c7-27ad-45f8-bb8c-e5f8828926cf","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2306.03341","last_updated":"2024-06-26T14:11:53Z","snapshot_observed_at":"2026-08-13T19:18:46.730073Z","submitted_at":"2023-06-06T01:26:53Z","title":"Inference-Time Intervention: Eliciting Truthful Answers from a Language Model","version":6},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2306.03341","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Inference- Time Intervention: Eliciting Truthful Answers from a Language Model, June 2024","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":18,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2306.03341","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:c247d83c9062304d0f9d23efe95b0de7d9f7b1f6376c318ffd39069e25f0b28f","observation_id":"408f0310-6202-47cb-9108-f1f9cc268c8c","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2209.07431","last_updated":"2022-09-21T14:30:36Z","snapshot_observed_at":"2026-08-16T16:32:13.907449Z","submitted_at":"2022-09-15T16:31:05Z","title":"Compositional generalization through abstract representations in human and artificial neural networks","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2209.07431","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Schultz, John D","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":19,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2209.07431","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:5bea8b02a24763258bf1a947a738e55556ecd724fdae24783ce29c3fc3a73d31","observation_id":"867a9b4b-68ac-4052-943b-2ed7bc5bcd44","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Jeffrey Johnston, and Stefano Fusi","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":20,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:774830ded0d24a3a6aba4214d6b6cc3317ba86129a3c72900b4f65c065ad5333","observation_id":"6ca20dc2-15d5-4370-95b7-cca9dc34a4ce","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2401.13558","last_updated":"2024-01-24T16:14:38Z","snapshot_observed_at":"2026-08-16T14:24:49.744227Z","submitted_at":"2024-01-24T16:14:38Z","title":"Task structure and nonlinearity jointly determine learned representational geometry","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2401.13558","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Lindsey, and Stefano Fusi","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":21,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2401.13558","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:2bbc1cd9ea975f292e33450d7a08556270881b81e6544ad6574c2763fc15fdf3","observation_id":"fe9e551c-316a-40ae-ac13-7bf498f29b6b","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Disentangling by Factorising","venue":null,"work_id":null,"year":2018},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":22,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:b0bcb3223318d72920566c72c18e29c44e25251254745f27eafbf4ff570ce140","observation_id":"668e7407-41d5-4093-900a-1e324272b670","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Jeffrey Johnston and Stefano Fusi","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":23,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:ab9b4b8d2de03784dd2bbbd8ded4c312cf64172aa1e2d13878be57003e371b4b","observation_id":"19519842-3938-408c-ad17-556f27a8339f","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Mickiewicz, James L","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":24,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:0556932c32e1fadfc1b520c6be257b9d1adb83d7e3ae4265e31cbd4e7d5e0440","observation_id":"2e203a87-be7e-4755-a0f6-65d68dd83f6c","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1312.6120","last_updated":"2014-02-19T17:26:57Z","snapshot_observed_at":"2026-08-14T23:50:45.351114Z","submitted_at":"2013-12-20T20:24:00Z","title":"Exact solutions to the nonlinear dynamics of learning in deep linear neural networks","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1312.6120","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Saxe, James L","venue":null,"work_id":null,"year":2014},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":25,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/1312.6120","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:3e92522e2789a78e9704ff634cd52aba611ad5ccae7936180fca663737d3a30a","observation_id":"4d1defce-f80b-47ad-9940-d68174449412","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1809.10374","last_updated":"2019-01-04T22:16:58Z","snapshot_observed_at":"2026-08-16T20:36:12.569927Z","submitted_at":"2018-09-27T06:47:58Z","title":"An analytic theory of generalization dynamics and transfer learning in deep linear networks","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1809.10374","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Lampinen and Surya Ganguli","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":26,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/1809.10374","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:7e31a538ec74ec896ac55ef311110db1118c207e9ed7c68204ca209522e7cef3","observation_id":"77a0de79-c247-41a5-9db8-f6e490ec4bf7","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":27,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:5b27580e5b3bb471c63a417e26b23e2fc62fc613703f4098d3d10f0ca92a1dd1","observation_id":"2accb746-4c48-4451-81dc-dd8b1a97f352","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2502.21009","last_updated":"2025-05-26T13:30:50Z","snapshot_observed_at":"2026-08-16T12:54:17.769856Z","submitted_at":"2025-02-28T12:52:11Z","title":"Position: Solve Layerwise Linear Models First to Understand Neural Dynamical Phenomena (Neural Collapse, Emergence, Lazy/Rich Regime, and Grokking)","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2502.21009","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":28,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2502.21009","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:52241589910d60b12fc45625d3cd51acc77598ccaeb79bc1c773d8dbf800abbc","observation_id":"138b986a-0070-410c-a1b1-413c811a4a4e","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2604.21691","last_updated":"2026-04-23T13:58:12Z","snapshot_observed_at":"2026-08-02T12:48:50.592713Z","submitted_at":"2026-04-23T13:58:12Z","title":"There Will Be a Scientific Theory of Deep Learning","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2604.21691","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Michaud, Berkan Ottlik, and Joseph Turnbull","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":29,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2604.21691","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:c2f213696e0662db5330ed94d33ea0c14f64f386aa58fa9c58032ab96b51d0fc","observation_id":"323c044c-2d29-4f78-868e-714ef08b1163","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Exact learning dynam- ics of deep linear networks with prior knowledge.Advances in Neural Information Processing Systems, 35:6615–6629, December 2022","venue":null,"work_id":null,"year":2022},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":30,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:82965c991fa9f6f929a486cd4872c390e8d6ac4acfeec94941ead1ecf84bcb8f","observation_id":"ae8e7888-1544-4781-bee7-2256cff180f8","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2409.14623","last_updated":"2025-03-04T11:18:33Z","snapshot_observed_at":"2026-08-16T13:16:26.554379Z","submitted_at":"2024-09-22T23:19:04Z","title":"From Lazy to Rich: Exact Learning Dynamics in Deep Linear Networks","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2409.14623","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":31,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2409.14623","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:3ad8560d09cbcb0124751b7693b816a2d83dbf304fe61987ad36dbe2d4ffb371","observation_id":"b0a81c82-badd-4a6b-829e-e1e2999aa7d3","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Korchinski, Dhruva Karkada, Yasaman Bahri, and Matthieu Wyart","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":32,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:2fbd3abaf24516e9330e204a50b9160ebab503c8059f2e1a2e02d20cf5365395","observation_id":"f6fffcc5-70af-4be1-a907-cbf987a3b59d","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":1995},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":33,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:44e2fe85563d8f489f097b1acbc5b52b396524de72b9f1eb8df59d3009474f25","observation_id":"7fa246b4-a1fc-4a84-a822-f595badc9917","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Du, Wei Hu, Zhiyuan Li, and Ruosong Wang","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":34,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:c4261da6053e26421954e6408359f06b63218ef6a79c3f26e84b99fc8dc0dec5","observation_id":"bc45e45a-1030-4d8f-82d0-a8cb4379dfaf","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1901.08584","last_updated":"2019-05-27T04:22:57Z","snapshot_observed_at":"2026-08-15T01:18:36.771990Z","submitted_at":"2019-01-24T18:58:16Z","title":"Fine-Grained Analysis of Optimization and Generalization for Overparameterized Two-Layer Neural Networks","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1901.08584","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":1901},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":35,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/1901.08584","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:565a160e3d4527113d5e2d116356a1d6bc8c614a9c8c6a42a6d0b23563696b50","observation_id":"a4789879-1abd-4478-8700-df0e4a44e4e8","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2504.18208","last_updated":"2025-07-21T14:02:53Z","snapshot_observed_at":"2026-08-18T04:03:20.363617Z","submitted_at":"2025-04-25T09:40:10Z","title":"Ultra-fast feature learning for the training of two-layer neural networks in the two-timescale regime","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2504.18208","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Ultra-fast feature learning for the training of two-layer neural networks in the two-timescale regime, July 2025","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":36,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2504.18208","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:4b34d3f620951b86ff8abb00abec75ab10084fc200fd20e61bf9a000de6d30e3","observation_id":"426ea73a-23d6-4685-9315-7ced04b2844a","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2304.09576","last_updated":"2023-10-25T14:13:57Z","snapshot_observed_at":"2026-08-16T15:39:19.017246Z","submitted_at":"2023-04-19T11:27:09Z","title":"Leveraging the two timescale regime to demonstrate convergence of neural networks","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2304.09576","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Leveraging the two timescale regime to demonstrate convergence of neural networks, October 2023","venue":null,"work_id":null,"year":2023},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":37,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2304.09576","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:476b20a471444e50296a799ae376f6ade1947e92b54eebfef6cdd0957d86249d","observation_id":"538507ff-581f-452b-9dbd-3eaa31527630","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1810.02281","last_updated":"2019-10-26T06:58:22Z","snapshot_observed_at":"2026-08-15T13:24:42.177517Z","submitted_at":"2018-10-04T15:53:32Z","title":"A Convergence Analysis of Gradient Descent for Deep Linear Neural Networks","version":3},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1810.02281","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"A Convergence Analysis of Gradi- ent Descent for Deep Linear Neural Networks, October 2019","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":38,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/1810.02281","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:5e19ee3b19aec61aaf2555de27603ffee63ef859eb9104b9511c3814e1c887c9","observation_id":"8f421669-c9b0-4bb9-96e3-7a41559a4b0c","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Kernel Methods for Deep Learning","venue":null,"work_id":null,"year":2009},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":39,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:b3453463b8446ecaf704f82730c7140379ca1374921eb6522e35ab5ae05fdc47","observation_id":"f4d49287-3728-41a2-821c-617f29b5b18c","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Computing with Infinite Networks","venue":null,"work_id":null,"year":1996},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":40,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:6c2016f98807d93a2f07b9f5a88fac7c8982e10edabcc85cf26811fdedcf0d91","observation_id":"484c21b6-4f43-48a3-b947-f0de95864923","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2508.10104","last_updated":"2025-08-13T18:00:55Z","snapshot_observed_at":"2026-08-18T01:07:23.737664Z","submitted_at":"2025-08-13T18:00:55Z","title":"DINOv3","version":1},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2508.10104","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":41,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2508.10104","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:9d16a7bff201795623dcfcd69d3000ae31c0e0265785440b97b84fc6d95dbd9b","observation_id":"270d1a44-ff3e-4b87-a98f-1b56fdbe7116","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"2403.08295","last_updated":"2024-04-16T12:52:47Z","snapshot_observed_at":"2026-08-03T03:29:01.959523Z","submitted_at":"2024-03-13T06:59:16Z","title":"Gemma: Open Models Based on Gemini Research and Technology","version":4},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"2403.08295","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":null,"venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":42,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/2403.08295","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:33849db69e6c869c224bbdbfb5902868cc4e7c9a27d7c3d4f9b47859f30a8594","observation_id":"1d88c27d-a04e-47d7-b716-2162c93801a9","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Majaj, Ha Hong, Ethan A","venue":null,"work_id":null,"year":2015},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":43,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:23d835cee2e46fe1e0140c88fc0046f97ac38e772b03507dd0b0808627cacc03","observation_id":"3af46104-04b0-44b1-a32d-d61afcbcd6b9","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":{"arxiv_id":"1901.09813","last_updated":"2019-05-11T19:02:42Z","snapshot_observed_at":"2026-08-14T17:24:48.311314Z","submitted_at":"2019-01-28T17:04:25Z","title":"Analogies Explained: Towards Understanding Word Embeddings","version":2},"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":"1901.09813","snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Analogies Explained: Towards Understanding Word Embeddings, May 2019","venue":null,"work_id":null,"year":2019},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":44,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"cited_paper":"/paper/1901.09813","citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:114954be3e9a101e090f5872cb1ebeb3932ed9d6ba52079aa739a93cf3d51397","observation_id":"ee37bed6-06b7-4bbd-8cb0-0b7e1a24bd64","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Simon, Yasaman Bahri, and Michael R","venue":null,"work_id":null,"year":2025},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":45,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:6811617f7ad810843ef466f61a33d6d49dc5291bce40d9863f9c374295240177","observation_id":"8d4e6d2c-cd01-498b-ae38-524cf0ebbf01","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Korchinski, Andres Nava, Matthieu Wyart, and Yasaman Bahri","venue":null,"work_id":null,"year":2026},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":46,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:46f02b09a1f9cc8914dec29a9afdca58e8205aa9a0960cf0195c5d0e30c1a306","observation_id":"6c3a31e5-8379-4b21-b5e3-1901ae827d18","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Separable nonlinear least squares: The variable projection method and its applications.Inverse Problems, 19(2):R1, February 2003","venue":null,"work_id":null,"year":2003},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":47,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:60fac80454921bc533b81ac381f28f72f61bf8424421a56960f063cd648c92b3","observation_id":"e983af7e-7745-4273-b3af-19362473c766","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Training Two-Layered Feedforward Networks With Variable Projection Method.IEEE Transactions on Neural Networks, 19(2):371–375, February 2008","venue":null,"work_id":null,"year":2008},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":48,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:d09af9f76c6477f3cb3b3cafcfe0a63afa982a46191657856a239e2f7d11c7f4","observation_id":"3221e9e8-2691-434a-8689-51d828adc212","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"abstract","venue":null,"work_id":null,"year":2024},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":49,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:fc01de3d9459477304c9f0451f379685fe4508a9bf154597bc5940798d861e2d","observation_id":"fba0e60b-0607-4333-bef3-df41bbe2f552","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}},{"citation":{"cited_paper":null,"cited_work":{"arxiv_id":null,"doi":null,"metadata_source":null,"pith_arxiv_id":null,"snapshot_observed_at":"2026-07-13T06:19:30.027337Z","title":"Each concept is instantiated by 80 ordered word pairs comprising common words","venue":null,"work_id":null,"year":null},"citing_paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction","version":1},"reference_index":50,"source":"pdf_text","source_observed_at":"2026-07-13T06:19:30.027337Z"},"links":{"citing_paper":"/paper/2607.08843"},"observation_digest":"sha256:dd3b93d496326db31494a1a3f9053cfcb33ccd232f0114f34a1979efffb38e9f","observation_id":"a94ce804-dbbd-418f-93b7-2f4c838c8e39","resolution":{"observed_at":"2026-07-13T06:19:30.027337Z","resolver_source":null,"status":"unresolved"},"standing_notice":{"events":[],"reason":"canonical_work_link_unavailable","source_receipts":[],"state":"unavailable"}}],"paper":{"arxiv_id":"2607.08843","last_updated":"2026-07-09T18:04:25Z","latest_version":1,"primary_category":"cs.LG","snapshot_observed_at":"2026-08-17T20:53:26.695918Z","submitted_at":"2026-07-09T18:04:25Z","title":"How are linear representations learned? Exact solutions to the dynamics of abstraction"},"reference_resolution":{"displayed":50,"state_counts":{"malformed_identifier":0,"metadata_mismatch":0,"parse_uncertain":0,"unresolved":50,"verified_exact":0,"verified_fuzzy":0},"total_outbound_references":50},"refusal":"A citation records a reference. It does not transfer a finding from one paper to another.","schema":"pith.paper-citation-record.v1","standing_sources":[{"observed_at":"2026-08-19T06:32:44.657259+00:00","source":"crossref"},{"observed_at":"2026-08-19T06:32:39.956319+00:00","source":"retraction_watch"}],"thesis":"As of 19 August 2026, this Paper Citation Record lists 50 of 50 outbound references and 0 inbound Pith citation observations for arXiv:2607.08843."}