{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:RFNBILVC2EHSUOK7FVN64TKJHJ","short_pith_number":"pith:RFNBILVC","schema_version":"1.0","canonical_sha256":"895a142ea2d10f2a395f2d5bee4d493a6b684c90b8956c4cbbedbf329f700e97","source":{"kind":"arxiv","id":"2210.13382","version":5},"attestation_state":"computed","paper":{"title":"Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Aspen K. Hopkins, David Bau, Fernanda Vi\\'egas, Hanspeter Pfister, Kenneth Li, Martin Wattenberg","submitted_at":"2022-10-24T16:29:55Z","abstract_excerpt":"Language models show a surprising range of capabilities, but the source of their apparent competence is unclear. Do these networks just memorize a collection of surface statistics, or do they rely on internal representations of the process that generates the sequences they see? We investigate this question by applying a variant of the GPT model to the task of predicting legal moves in a simple board game, Othello. Although the network has no a priori knowledge of the game or its rules, we uncover evidence of an emergent nonlinear internal representation of the board state. Interventional exper"},"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":"2210.13382","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-10-24T16:29:55Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"f28b004cfc11743661a0bb7e050f38cb6e864ba7d38abb1908ae29165616bca2","abstract_canon_sha256":"c6c1de5bbeed97b2853b475aff08b4a1efcd4a4ce0f29a2d5dac247238ae6ec3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:37:03.437785Z","signature_b64":"NC0rW/LnXnpF3aV5xCCR9iAD6BifgY91mLPuHSuJDMNMR+XGTfubZgZVS0O7qWWTBvqvUf4HxNw+Zw8zqYsZAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"895a142ea2d10f2a395f2d5bee4d493a6b684c90b8956c4cbbedbf329f700e97","last_reissued_at":"2026-07-05T08:37:03.437347Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:37:03.437347Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Aspen K. Hopkins, David Bau, Fernanda Vi\\'egas, Hanspeter Pfister, Kenneth Li, Martin Wattenberg","submitted_at":"2022-10-24T16:29:55Z","abstract_excerpt":"Language models show a surprising range of capabilities, but the source of their apparent competence is unclear. Do these networks just memorize a collection of surface statistics, or do they rely on internal representations of the process that generates the sequences they see? We investigate this question by applying a variant of the GPT model to the task of predicting legal moves in a simple board game, Othello. Although the network has no a priori knowledge of the game or its rules, we uncover evidence of an emergent nonlinear internal representation of the board state. Interventional exper"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.13382","kind":"arxiv","version":5},"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/2210.13382/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":"2210.13382","created_at":"2026-07-05T08:37:03.437405+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.13382v5","created_at":"2026-07-05T08:37:03.437405+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.13382","created_at":"2026-07-05T08:37:03.437405+00:00"},{"alias_kind":"pith_short_12","alias_value":"RFNBILVC2EHS","created_at":"2026-07-05T08:37:03.437405+00:00"},{"alias_kind":"pith_short_16","alias_value":"RFNBILVC2EHSUOK7","created_at":"2026-07-05T08:37:03.437405+00:00"},{"alias_kind":"pith_short_8","alias_value":"RFNBILVC","created_at":"2026-07-05T08:37:03.437405+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":18,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24669","citing_title":"LaGO: Latent Action Guidance for Online Reinforcement Learning","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26523","citing_title":"Radical AI Interpretability","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08644","citing_title":"A retrieval conditioned rebinding circuit for dynamic entity tracking in large language models","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05106","citing_title":"Arithmetic Pedagogy for Language Models","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03685","citing_title":"A Close Look At World Model Recovery In Supervised Fine-Tuned LLM Planners","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23972","citing_title":"Why We Need World Models for AGI: Where LLMs Fail and How World Models May Outperform","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2606.20600","citing_title":"The New Associationism: Lessons from Deep Learning","ref_index":90,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25384","citing_title":"GeoMathCode: Understanding Interleaved Math-Code Reasoning for Geometry Problem Solving","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29358","citing_title":"Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20299","citing_title":"Mechanisms of Misgeneralization in Physical Sequence Modeling","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16665","citing_title":"In-context learning enables continental-scale subsurface temperature prediction from sparse local observations","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17510","citing_title":"Scale-Dependent Collective Adaptation in Self-Amending LLM Societies: A Cross-Family Study of Emergent Governance","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2303.08112","citing_title":"Eliciting Latent Predictions from Transformers with the Tuned Lens","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09967","citing_title":"Tensor Product Representation Probes Reveal Shared Structure Across Linear Directions","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09693","citing_title":"Do multimodal models imagine electric sheep?","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2604.25905","citing_title":"A paradox of AI fluency","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05593","citing_title":"Causal Probing for Internal Visual Representations in Multimodal Large Language Models","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19052","citing_title":"Cell-Based Representation of Relational Binding in Language Models","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ","json":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ.json","graph_json":"https://pith.science/api/pith-number/RFNBILVC2EHSUOK7FVN64TKJHJ/graph.json","events_json":"https://pith.science/api/pith-number/RFNBILVC2EHSUOK7FVN64TKJHJ/events.json","paper":"https://pith.science/paper/RFNBILVC"},"agent_actions":{"view_html":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ","download_json":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ.json","view_paper":"https://pith.science/paper/RFNBILVC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.13382&json=true","fetch_graph":"https://pith.science/api/pith-number/RFNBILVC2EHSUOK7FVN64TKJHJ/graph.json","fetch_events":"https://pith.science/api/pith-number/RFNBILVC2EHSUOK7FVN64TKJHJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ/action/storage_attestation","attest_author":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ/action/author_attestation","sign_citation":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ/action/citation_signature","submit_replication":"https://pith.science/pith/RFNBILVC2EHSUOK7FVN64TKJHJ/action/replication_record"}},"created_at":"2026-07-05T08:37:03.437405+00:00","updated_at":"2026-07-05T08:37:03.437405+00:00"}