{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:4SJQMRSFWVSHPM257AYDAQXEGM","short_pith_number":"pith:4SJQMRSF","canonical_record":{"source":{"id":"2605.14125","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-13T21:21:10Z","cross_cats_sorted":[],"title_canon_sha256":"4c58f5d0e3aa47dca3aebdc9c6f93c67e35dd7ebe0f26bbc067bd7c18be014de","abstract_canon_sha256":"bbd664110450ecd094b98f51b6a32875c3edd13bb542e6309805f7eab8620dba"},"schema_version":"1.0"},"canonical_sha256":"e493064645b56477b35df8303042e4330787816c638d00ace3bd6ef331af8b11","source":{"kind":"arxiv","id":"2605.14125","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14125","created_at":"2026-05-17T23:39:11Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14125v1","created_at":"2026-05-17T23:39:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14125","created_at":"2026-05-17T23:39:11Z"},{"alias_kind":"pith_short_12","alias_value":"4SJQMRSFWVSH","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"4SJQMRSFWVSHPM25","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"4SJQMRSF","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:4SJQMRSFWVSHPM257AYDAQXEGM","target":"record","payload":{"canonical_record":{"source":{"id":"2605.14125","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-13T21:21:10Z","cross_cats_sorted":[],"title_canon_sha256":"4c58f5d0e3aa47dca3aebdc9c6f93c67e35dd7ebe0f26bbc067bd7c18be014de","abstract_canon_sha256":"bbd664110450ecd094b98f51b6a32875c3edd13bb542e6309805f7eab8620dba"},"schema_version":"1.0"},"canonical_sha256":"e493064645b56477b35df8303042e4330787816c638d00ace3bd6ef331af8b11","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:11.857132Z","signature_b64":"4amISc0QbU5TNVNaf3luJUd3qSJkC9yuAdnEpiAPzb/orkPm/LYnJu1eMxj97csHw8IkvrMp3yNFyPI2TAywCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e493064645b56477b35df8303042e4330787816c638d00ace3bd6ef331af8b11","last_reissued_at":"2026-05-17T23:39:11.856493Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:11.856493Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.14125","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:11Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HSTf9fLCcDo39Dk5Ji7uQn6rccfXe90axlEGdq2nZgTiOD1Np7ETC2yqVMxvYxybQr4VedE7YETU1tTMENSUBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T23:20:38.396813Z"},"content_sha256":"c85370d062a17b3028cf6abb3b0a59d0ab96c8dfb1cc914e73dcb1f66e92dac6","schema_version":"1.0","event_id":"sha256:c85370d062a17b3028cf6abb3b0a59d0ab96c8dfb1cc914e73dcb1f66e92dac6"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:4SJQMRSFWVSHPM257AYDAQXEGM","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Polar probe linearly decodes semantic structures from LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Large language models encode the existence and type of semantic relations as distance and direction between embeddings in a linear subspace of their activations.","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jean-R\\'emi King, Pablo J. Diego-Sim\\'on, Pierre Orhan, Yair Lakretz","submitted_at":"2026-05-13T21:21:10Z","abstract_excerpt":"How do artificial neural networks bind concepts to form complex semantic structures? Here, we propose a simple neural code, whereby the existence and the type of relations between entities are represented by the distance and the direction between their embeddings, respectively. We test this hypothesis in a variety of Large Language Models (LLMs), each input with natural-language descriptions of minimalist tasks from five different domains: arithmetic, visual scenes, family trees, metro maps and social interactions. Results show that the true semantic structures can be linearly recovered with a"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"the true semantic structures can be linearly recovered with a Polar Probe targeting a subspace of LLMs' layer activations","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"that the distance-direction geometry is the actual causal mechanism used by the model rather than a convenient post-hoc linear fit that happens to correlate with task performance","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LLMs bind semantic relations via polar geometry in embeddings, linearly decodable from middle layers with generalization to new items.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Large language models encode the existence and type of semantic relations as distance and direction between embeddings in a linear subspace of their activations.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"15145f4ee4a4a75d059588396ddf88311f6b94c9f4bd73aba78af283a80d5543"},"source":{"id":"2605.14125","kind":"arxiv","version":1},"verdict":{"id":"3ad1e17e-5e49-4531-8d07-586c370591e3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T04:56:34.852465Z","strongest_claim":"the true semantic structures can be linearly recovered with a Polar Probe targeting a subspace of LLMs' layer activations","one_line_summary":"LLMs bind semantic relations via polar geometry in embeddings, linearly decodable from middle layers with generalization to new items.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"that the distance-direction geometry is the actual causal mechanism used by the model rather than a convenient post-hoc linear fit that happens to correlate with task performance","pith_extraction_headline":"Large language models encode the existence and type of semantic relations as distance and direction between embeddings in a linear subspace of their activations."},"references":{"count":112,"sample":[{"doi":"","year":2017,"title":"Understanding intermediate layers using linear classifier probes, 2017","work_id":"9a6797e3-8303-4253-8a12-d138400e0736","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Pythia: a suite for analyzing large language models across training and scaling","work_id":"ac8fbbb8-291b-4844-84a1-34d3bf2a9a70","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Fast differentiable sorting and ranking","work_id":"cfcdb5d4-d4b2-4230-9131-1637b81f4ec4","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1901,"title":"Language models are few-shot learners","work_id":"33f1f5a3-ad49-41fa-b313-857dabbce8e9","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Diego-Simon, St \\'e phane d'Ascoli, Emmanuel Chemla, Yair Lakretz, and Jean-Remi King","work_id":"dfd927f4-f5fa-4e49-a993-6960160f2f68","ref_index":8,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":112,"snapshot_sha256":"7640a76a3297997726cb3b2cb48d8ed95864a548fdcc4c669214b1e4f834201b","internal_anchors":2},"formal_canon":{"evidence_count":3,"snapshot_sha256":"0c5ee8b3dda2b6c106c0312ba7bfae38b73465ba6b0c496297fcb46c8e3d0939"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"3ad1e17e-5e49-4531-8d07-586c370591e3"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:11Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"dBK2ac8QCX3HZfJmxSvO0ENyJds9WdqWzi0bf6FvTvGZReU2Pp70oXDz9I0UjFW29EOcjxxT31GOrwEO4+7/BA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-27T23:20:38.397662Z"},"content_sha256":"81f1675e3d97126ddb617a8d674fd2985c6b8c7e182f9970a2a5be968eee8b25","schema_version":"1.0","event_id":"sha256:81f1675e3d97126ddb617a8d674fd2985c6b8c7e182f9970a2a5be968eee8b25"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/4SJQMRSFWVSHPM257AYDAQXEGM/bundle.json","state_url":"https://pith.science/pith/4SJQMRSFWVSHPM257AYDAQXEGM/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/4SJQMRSFWVSHPM257AYDAQXEGM/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":"2026-05-27T23:20:38Z","links":{"resolver":"https://pith.science/pith/4SJQMRSFWVSHPM257AYDAQXEGM","bundle":"https://pith.science/pith/4SJQMRSFWVSHPM257AYDAQXEGM/bundle.json","state":"https://pith.science/pith/4SJQMRSFWVSHPM257AYDAQXEGM/state.json","well_known_bundle":"https://pith.science/.well-known/pith/4SJQMRSFWVSHPM257AYDAQXEGM/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:4SJQMRSFWVSHPM257AYDAQXEGM","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":"bbd664110450ecd094b98f51b6a32875c3edd13bb542e6309805f7eab8620dba","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-13T21:21:10Z","title_canon_sha256":"4c58f5d0e3aa47dca3aebdc9c6f93c67e35dd7ebe0f26bbc067bd7c18be014de"},"schema_version":"1.0","source":{"id":"2605.14125","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14125","created_at":"2026-05-17T23:39:11Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14125v1","created_at":"2026-05-17T23:39:11Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14125","created_at":"2026-05-17T23:39:11Z"},{"alias_kind":"pith_short_12","alias_value":"4SJQMRSFWVSH","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"4SJQMRSFWVSHPM25","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"4SJQMRSF","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:81f1675e3d97126ddb617a8d674fd2985c6b8c7e182f9970a2a5be968eee8b25","target":"graph","created_at":"2026-05-17T23:39:11Z","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":"the true semantic structures can be linearly recovered with a Polar Probe targeting a subspace of LLMs' layer activations"},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"that the distance-direction geometry is the actual causal mechanism used by the model rather than a convenient post-hoc linear fit that happens to correlate with task performance"},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"LLMs bind semantic relations via polar geometry in embeddings, linearly decodable from middle layers with generalization to new items."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Large language models encode the existence and type of semantic relations as distance and direction between embeddings in a linear subspace of their activations."}],"snapshot_sha256":"15145f4ee4a4a75d059588396ddf88311f6b94c9f4bd73aba78af283a80d5543"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"0c5ee8b3dda2b6c106c0312ba7bfae38b73465ba6b0c496297fcb46c8e3d0939"},"paper":{"abstract_excerpt":"How do artificial neural networks bind concepts to form complex semantic structures? Here, we propose a simple neural code, whereby the existence and the type of relations between entities are represented by the distance and the direction between their embeddings, respectively. We test this hypothesis in a variety of Large Language Models (LLMs), each input with natural-language descriptions of minimalist tasks from five different domains: arithmetic, visual scenes, family trees, metro maps and social interactions. Results show that the true semantic structures can be linearly recovered with a","authors_text":"Jean-R\\'emi King, Pablo J. Diego-Sim\\'on, Pierre Orhan, Yair Lakretz","cross_cats":[],"headline":"Large language models encode the existence and type of semantic relations as distance and direction between embeddings in a linear subspace of their activations.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-13T21:21:10Z","title":"Polar probe linearly decodes semantic structures from LLMs"},"references":{"count":112,"internal_anchors":2,"resolved_work":112,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Understanding intermediate layers using linear classifier probes, 2017","work_id":"9a6797e3-8303-4253-8a12-d138400e0736","year":2017},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Pythia: a suite for analyzing large language models across training and scaling","work_id":"ac8fbbb8-291b-4844-84a1-34d3bf2a9a70","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Fast differentiable sorting and ranking","work_id":"cfcdb5d4-d4b2-4230-9131-1637b81f4ec4","year":2020},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Language models are few-shot learners","work_id":"33f1f5a3-ad49-41fa-b313-857dabbce8e9","year":1901},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":8,"title":"Diego-Simon, St \\'e phane d'Ascoli, Emmanuel Chemla, Yair Lakretz, and Jean-Remi King","work_id":"dfd927f4-f5fa-4e49-a993-6960160f2f68","year":2024}],"snapshot_sha256":"7640a76a3297997726cb3b2cb48d8ed95864a548fdcc4c669214b1e4f834201b"},"source":{"id":"2605.14125","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T04:56:34.852465Z","id":"3ad1e17e-5e49-4531-8d07-586c370591e3","model_set":{"reader":"grok-4.3"},"one_line_summary":"LLMs bind semantic relations via polar geometry in embeddings, linearly decodable from middle layers with generalization to new items.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Large language models encode the existence and type of semantic relations as distance and direction between embeddings in a linear subspace of their activations.","strongest_claim":"the true semantic structures can be linearly recovered with a Polar Probe targeting a subspace of LLMs' layer activations","weakest_assumption":"that the distance-direction geometry is the actual causal mechanism used by the model rather than a convenient post-hoc linear fit that happens to correlate with task performance"}},"verdict_id":"3ad1e17e-5e49-4531-8d07-586c370591e3"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c85370d062a17b3028cf6abb3b0a59d0ab96c8dfb1cc914e73dcb1f66e92dac6","target":"record","created_at":"2026-05-17T23:39:11Z","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":"bbd664110450ecd094b98f51b6a32875c3edd13bb542e6309805f7eab8620dba","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-13T21:21:10Z","title_canon_sha256":"4c58f5d0e3aa47dca3aebdc9c6f93c67e35dd7ebe0f26bbc067bd7c18be014de"},"schema_version":"1.0","source":{"id":"2605.14125","kind":"arxiv","version":1}},"canonical_sha256":"e493064645b56477b35df8303042e4330787816c638d00ace3bd6ef331af8b11","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e493064645b56477b35df8303042e4330787816c638d00ace3bd6ef331af8b11","first_computed_at":"2026-05-17T23:39:11.856493Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:11.856493Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"4amISc0QbU5TNVNaf3luJUd3qSJkC9yuAdnEpiAPzb/orkPm/LYnJu1eMxj97csHw8IkvrMp3yNFyPI2TAywCA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:11.857132Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14125","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c85370d062a17b3028cf6abb3b0a59d0ab96c8dfb1cc914e73dcb1f66e92dac6","sha256:81f1675e3d97126ddb617a8d674fd2985c6b8c7e182f9970a2a5be968eee8b25"],"state_sha256":"c453ca163ce52d6c5e8698d5c2c6f265461d1f6bb86decb3135a3fdc79cdd46e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GXUp/eCkg1VA3MMYWOB4McYjkxqJAjYtONv+cFScSyTRKT6zx0McSKkVWSnEEx/fdiq4UcoNPYt37p9DL+E6Dw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-27T23:20:38.402376Z","bundle_sha256":"43c134016a9d8cd7f86a8426c148e67ec2b260c125c7866e6dce872885103a25"}}