{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:O5FGSQF52JYQJG6M4EJUGWR4RU","short_pith_number":"pith:O5FGSQF5","canonical_record":{"source":{"id":"2311.03658","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-07T01:59:11Z","cross_cats_sorted":["cs.AI","cs.LG","stat.ML"],"title_canon_sha256":"4f38a4423afec1c2192b83aca612444daf27dedf0b6ae368025085f67b69be7f","abstract_canon_sha256":"52fc5acc1032b265edd952df1098bf1b816a082f7237490d8038ab7862d67fac"},"schema_version":"1.0"},"canonical_sha256":"774a6940bdd271049bcce113435a3c8d3c31947ca6c396b22ef91cc32f9ea2f9","source":{"kind":"arxiv","id":"2311.03658","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2311.03658","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"arxiv_version","alias_value":"2311.03658v2","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.03658","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"pith_short_12","alias_value":"O5FGSQF52JYQ","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"pith_short_16","alias_value":"O5FGSQF52JYQJG6M","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"pith_short_8","alias_value":"O5FGSQF5","created_at":"2026-05-20T00:00:14Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:O5FGSQF52JYQJG6M4EJUGWR4RU","target":"record","payload":{"canonical_record":{"source":{"id":"2311.03658","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-07T01:59:11Z","cross_cats_sorted":["cs.AI","cs.LG","stat.ML"],"title_canon_sha256":"4f38a4423afec1c2192b83aca612444daf27dedf0b6ae368025085f67b69be7f","abstract_canon_sha256":"52fc5acc1032b265edd952df1098bf1b816a082f7237490d8038ab7862d67fac"},"schema_version":"1.0"},"canonical_sha256":"774a6940bdd271049bcce113435a3c8d3c31947ca6c396b22ef91cc32f9ea2f9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:14.504209Z","signature_b64":"wBApiqTmHRDwzDH+9ZvDbgH8kjym8zG8SL64vMY0eup2GmR6En7+MR5Hs7KGsmE0VbxVe/YgVwe0akSABWtICw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"774a6940bdd271049bcce113435a3c8d3c31947ca6c396b22ef91cc32f9ea2f9","last_reissued_at":"2026-05-20T00:00:14.503329Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:14.503329Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2311.03658","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+h8W+32PNXc31AIiEYmqyPg/Oww6zAE9lsOgoMwDrlsKT2k29EabqsKolimSV+ns9Y9Nl28qlt0iIU857QyEDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-01T01:06:44.758696Z"},"content_sha256":"3e9d2516bb0d8c4423595c8caf16791044de2ea996ce68e634fdccbe0f7b63b0","schema_version":"1.0","event_id":"sha256:3e9d2516bb0d8c4423595c8caf16791044de2ea996ce68e634fdccbe0f7b63b0"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:O5FGSQF52JYQJG6M4EJUGWR4RU","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"The Linear Representation Hypothesis and the Geometry of Large Language Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"High-level concepts in large language models are linear directions under a causal inner product built from counterfactual pairs.","cross_cats":["cs.AI","cs.LG","stat.ML"],"primary_cat":"cs.CL","authors_text":"Kiho Park, Victor Veitch, Yo Joong Choe","submitted_at":"2023-11-07T01:59:11Z","abstract_excerpt":"Informally, the 'linear representation hypothesis' is the idea that high-level concepts are represented linearly as directions in some representation space. In this paper, we address two closely related questions: What does \"linear representation\" actually mean? And, how do we make sense of geometric notions (e.g., cosine similarity or projection) in the representation space? To answer these, we use the language of counterfactuals to give two formalizations of \"linear representation\", one in the output (word) representation space, and one in the input (sentence) space. We then prove these conn"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Using this causal inner product, we show how to unify all notions of linear representation. In particular, this allows the construction of probes and steering vectors using counterfactual pairs.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that the identified non-Euclidean inner product respects language structure in the precise sense required to unify probing and steering, and that counterfactual pairs can be reliably constructed or approximated in the model.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Linear representations of high-level concepts in LLMs are formalized via counterfactuals in input and output spaces, unified under a causal inner product that enables consistent probing and steering.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"High-level concepts in large language models are linear directions under a causal inner product built from counterfactual pairs.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c93988df41b25e5342f6f134c613f3024f88c1bf67e49eb4bbd03da16ec2c409"},"source":{"id":"2311.03658","kind":"arxiv","version":2},"verdict":{"id":"86afa7eb-3a40-40d5-86e8-56e7949c4e8a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T21:37:51.886037Z","strongest_claim":"Using this causal inner product, we show how to unify all notions of linear representation. In particular, this allows the construction of probes and steering vectors using counterfactual pairs.","one_line_summary":"Linear representations of high-level concepts in LLMs are formalized via counterfactuals in input and output spaces, unified under a causal inner product that enables consistent probing and steering.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that the identified non-Euclidean inner product respects language structure in the precise sense required to unify probing and steering, and that counterfactual pairs can be reliably constructed or approximated in the model.","pith_extraction_headline":"High-level concepts in large language models are linear directions under a causal inner product built from counterfactual pairs."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2311.03658/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":30,"sample":[{"doi":"10.18653/v1/k16-1002","year":2022,"title":"doi: 10.18653/v1/K16-1002","work_id":"54f0083e-6c5a-47d8-9d80-a6ed0da3f854","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"Word embed- dings, analogies, and machine learning: Beyond king - man + woman = queen","work_id":"884ff28a-7e39-45f3-a903-eb6950c0b799","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Toy Models of Superposition","work_id":"43875dbe-bc2d-4ab5-af63-744411533ff7","ref_index":3,"cited_arxiv_id":"2209.10652","is_internal_anchor":true},{"doi":"","year":2019,"title":"How contextual are contextualized word rep- resentations? Comparing the geometry of BERT, ELMo, and GPT-2 embeddings","work_id":"d68d8fa1-a3bb-459b-ab72-d52f506d7a78","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.18653/v1/2020.conll-1.29","year":2020,"title":"doi: 10.18653/v1/2020.conll-1.29","work_id":"1eb9b7d9-cd8a-45e6-9636-c1dab1d9f4d5","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":30,"snapshot_sha256":"bd1b684c9f9c68682f6f9b55a65b560783157ab45073589cc574a93476b78276","internal_anchors":8},"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"},"verdict_id":"86afa7eb-3a40-40d5-86e8-56e7949c4e8a"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:00:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"yMbhoKSn40kTHphwCjztISEwWH4PaoXIEcsqPgtPKytfVoQS3dsxkaJzgC3ZTSuBHUV0cr5i9HGr3v2PAVhfBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-01T01:06:44.759866Z"},"content_sha256":"6a175258c50227696cdf4c2057ed858558db1060d94e9dcba316748c2cd9d92b","schema_version":"1.0","event_id":"sha256:6a175258c50227696cdf4c2057ed858558db1060d94e9dcba316748c2cd9d92b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/O5FGSQF52JYQJG6M4EJUGWR4RU/bundle.json","state_url":"https://pith.science/pith/O5FGSQF52JYQJG6M4EJUGWR4RU/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/O5FGSQF52JYQJG6M4EJUGWR4RU/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-01T01:06:44Z","links":{"resolver":"https://pith.science/pith/O5FGSQF52JYQJG6M4EJUGWR4RU","bundle":"https://pith.science/pith/O5FGSQF52JYQJG6M4EJUGWR4RU/bundle.json","state":"https://pith.science/pith/O5FGSQF52JYQJG6M4EJUGWR4RU/state.json","well_known_bundle":"https://pith.science/.well-known/pith/O5FGSQF52JYQJG6M4EJUGWR4RU/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:O5FGSQF52JYQJG6M4EJUGWR4RU","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"52fc5acc1032b265edd952df1098bf1b816a082f7237490d8038ab7862d67fac","cross_cats_sorted":["cs.AI","cs.LG","stat.ML"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-07T01:59:11Z","title_canon_sha256":"4f38a4423afec1c2192b83aca612444daf27dedf0b6ae368025085f67b69be7f"},"schema_version":"1.0","source":{"id":"2311.03658","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2311.03658","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"arxiv_version","alias_value":"2311.03658v2","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.03658","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"pith_short_12","alias_value":"O5FGSQF52JYQ","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"pith_short_16","alias_value":"O5FGSQF52JYQJG6M","created_at":"2026-05-20T00:00:14Z"},{"alias_kind":"pith_short_8","alias_value":"O5FGSQF5","created_at":"2026-05-20T00:00:14Z"}],"graph_snapshots":[{"event_id":"sha256:6a175258c50227696cdf4c2057ed858558db1060d94e9dcba316748c2cd9d92b","target":"graph","created_at":"2026-05-20T00:00:14Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Using this causal inner product, we show how to unify all notions of linear representation. In particular, this allows the construction of probes and steering vectors using counterfactual pairs."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The assumption that the identified non-Euclidean inner product respects language structure in the precise sense required to unify probing and steering, and that counterfactual pairs can be reliably constructed or approximated in the model."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Linear representations of high-level concepts in LLMs are formalized via counterfactuals in input and output spaces, unified under a causal inner product that enables consistent probing and steering."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"High-level concepts in large language models are linear directions under a causal inner product built from counterfactual pairs."}],"snapshot_sha256":"c93988df41b25e5342f6f134c613f3024f88c1bf67e49eb4bbd03da16ec2c409"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2311.03658/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Informally, the 'linear representation hypothesis' is the idea that high-level concepts are represented linearly as directions in some representation space. In this paper, we address two closely related questions: What does \"linear representation\" actually mean? And, how do we make sense of geometric notions (e.g., cosine similarity or projection) in the representation space? To answer these, we use the language of counterfactuals to give two formalizations of \"linear representation\", one in the output (word) representation space, and one in the input (sentence) space. We then prove these conn","authors_text":"Kiho Park, Victor Veitch, Yo Joong Choe","cross_cats":["cs.AI","cs.LG","stat.ML"],"headline":"High-level concepts in large language models are linear directions under a causal inner product built from counterfactual pairs.","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-07T01:59:11Z","title":"The Linear Representation Hypothesis and the Geometry of Large Language Models"},"references":{"count":30,"internal_anchors":8,"resolved_work":30,"sample":[{"cited_arxiv_id":"","doi":"10.18653/v1/k16-1002","is_internal_anchor":false,"ref_index":1,"title":"doi: 10.18653/v1/K16-1002","work_id":"54f0083e-6c5a-47d8-9d80-a6ed0da3f854","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Word embed- dings, analogies, and machine learning: Beyond king - man + woman = queen","work_id":"884ff28a-7e39-45f3-a903-eb6950c0b799","year":2016},{"cited_arxiv_id":"2209.10652","doi":"","is_internal_anchor":true,"ref_index":3,"title":"Toy Models of Superposition","work_id":"43875dbe-bc2d-4ab5-af63-744411533ff7","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"How contextual are contextualized word rep- resentations? Comparing the geometry of BERT, ELMo, and GPT-2 embeddings","work_id":"d68d8fa1-a3bb-459b-ab72-d52f506d7a78","year":2019},{"cited_arxiv_id":"","doi":"10.18653/v1/2020.conll-1.29","is_internal_anchor":false,"ref_index":5,"title":"doi: 10.18653/v1/2020.conll-1.29","work_id":"1eb9b7d9-cd8a-45e6-9636-c1dab1d9f4d5","year":2020}],"snapshot_sha256":"bd1b684c9f9c68682f6f9b55a65b560783157ab45073589cc574a93476b78276"},"source":{"id":"2311.03658","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-11T21:37:51.886037Z","id":"86afa7eb-3a40-40d5-86e8-56e7949c4e8a","model_set":{"reader":"grok-4.3"},"one_line_summary":"Linear representations of high-level concepts in LLMs are formalized via counterfactuals in input and output spaces, unified under a causal inner product that enables consistent probing and steering.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"High-level concepts in large language models are linear directions under a causal inner product built from counterfactual pairs.","strongest_claim":"Using this causal inner product, we show how to unify all notions of linear representation. In particular, this allows the construction of probes and steering vectors using counterfactual pairs.","weakest_assumption":"The assumption that the identified non-Euclidean inner product respects language structure in the precise sense required to unify probing and steering, and that counterfactual pairs can be reliably constructed or approximated in the model."}},"verdict_id":"86afa7eb-3a40-40d5-86e8-56e7949c4e8a"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:3e9d2516bb0d8c4423595c8caf16791044de2ea996ce68e634fdccbe0f7b63b0","target":"record","created_at":"2026-05-20T00:00:14Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"52fc5acc1032b265edd952df1098bf1b816a082f7237490d8038ab7862d67fac","cross_cats_sorted":["cs.AI","cs.LG","stat.ML"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-07T01:59:11Z","title_canon_sha256":"4f38a4423afec1c2192b83aca612444daf27dedf0b6ae368025085f67b69be7f"},"schema_version":"1.0","source":{"id":"2311.03658","kind":"arxiv","version":2}},"canonical_sha256":"774a6940bdd271049bcce113435a3c8d3c31947ca6c396b22ef91cc32f9ea2f9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"774a6940bdd271049bcce113435a3c8d3c31947ca6c396b22ef91cc32f9ea2f9","first_computed_at":"2026-05-20T00:00:14.503329Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:00:14.503329Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"wBApiqTmHRDwzDH+9ZvDbgH8kjym8zG8SL64vMY0eup2GmR6En7+MR5Hs7KGsmE0VbxVe/YgVwe0akSABWtICw==","signature_status":"signed_v1","signed_at":"2026-05-20T00:00:14.504209Z","signed_message":"canonical_sha256_bytes"},"source_id":"2311.03658","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3e9d2516bb0d8c4423595c8caf16791044de2ea996ce68e634fdccbe0f7b63b0","sha256:6a175258c50227696cdf4c2057ed858558db1060d94e9dcba316748c2cd9d92b"],"state_sha256":"f25222be723203dba3d5f3524dea37c7ca13722ad94e1d1fd03db12a36cf03c7"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"FvY5+atM1RfZt6mZ+sBv0agUx8C8iKlScv7gr4eOpbB80iuzUQo7eF8xK3kR4Uws+QsqE9Pr//lYwhFHbubLAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-01T01:06:44.764955Z","bundle_sha256":"ad39ad10de6722a45fc6e78be91b001c7854ed9c88411793bbc8fc98f74336ed"}}