{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:E3HL3C2TDH7CBKJ2KFCNA5LS6Y","short_pith_number":"pith:E3HL3C2T","canonical_record":{"source":{"id":"2403.03867","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-06T17:17:36Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"d44a0a341a6311250ee246d415ab9511beb21683367e7daf3a24c9f91b0bcdbd","abstract_canon_sha256":"13b1b26fb2523bc99d6cffa95c0dabef853cf9538037aefa96f87f41d6b9775c"},"schema_version":"1.0"},"canonical_sha256":"26cebd8b5319fe20a93a5144d07572f60856e0311cf091794bffb63cb1603877","source":{"kind":"arxiv","id":"2403.03867","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.03867","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"arxiv_version","alias_value":"2403.03867v1","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.03867","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"pith_short_12","alias_value":"E3HL3C2TDH7C","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"pith_short_16","alias_value":"E3HL3C2TDH7CBKJ2","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"pith_short_8","alias_value":"E3HL3C2T","created_at":"2026-07-05T07:53:02Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:E3HL3C2TDH7CBKJ2KFCNA5LS6Y","target":"record","payload":{"canonical_record":{"source":{"id":"2403.03867","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-06T17:17:36Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"d44a0a341a6311250ee246d415ab9511beb21683367e7daf3a24c9f91b0bcdbd","abstract_canon_sha256":"13b1b26fb2523bc99d6cffa95c0dabef853cf9538037aefa96f87f41d6b9775c"},"schema_version":"1.0"},"canonical_sha256":"26cebd8b5319fe20a93a5144d07572f60856e0311cf091794bffb63cb1603877","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:53:02.500220Z","signature_b64":"qR923socxKAMZ1+FDsiWu1aPjp/KlIsWHAeXgJTNSD1WQVAI6Ckzl1CqYiijgW+2RSM1k+Jjp3q75T5A5S4sBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"26cebd8b5319fe20a93a5144d07572f60856e0311cf091794bffb63cb1603877","last_reissued_at":"2026-07-05T07:53:02.499748Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:53:02.499748Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2403.03867","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":"2026-07-05T07:53:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BaZ8J6vzT8iKk0CqNxAglLBil/vpSnWm/jojcOF6YpCy/H0qXcNUXwGDKnw5e8a12eqIlXLkK28hHB1yDnIfDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T12:18:26.209765Z"},"content_sha256":"ec85e0b2fb7e131991e29d71dc3652caf0b5bf93ce95ca443b33b78d4003c039","schema_version":"1.0","event_id":"sha256:ec85e0b2fb7e131991e29d71dc3652caf0b5bf93ce95ca443b33b78d4003c039"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:E3HL3C2TDH7CBKJ2KFCNA5LS6Y","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"On the Origins of Linear Representations in Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.CL","authors_text":"Bryon Aragam, Goutham Rajendran, Pradeep Ravikumar, Victor Veitch, Yibo Jiang","submitted_at":"2024-03-06T17:17:36Z","abstract_excerpt":"Recent works have argued that high-level semantic concepts are encoded \"linearly\" in the representation space of large language models. In this work, we study the origins of such linear representations. To that end, we introduce a simple latent variable model to abstract and formalize the concept dynamics of the next token prediction. We use this formalism to show that the next token prediction objective (softmax with cross-entropy) and the implicit bias of gradient descent together promote the linear representation of concepts. Experiments show that linear representations emerge when learning"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.03867","kind":"arxiv","version":1},"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/2403.03867/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T07:53:02Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"XB/y54gJ9s6ogcS/NEZjKlkqOxt4nedu89l24hUiVENAuepEln9vFy3HK/YzFPKnFnvNKNhOvTZ9DJc97J2HAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T12:18:26.210285Z"},"content_sha256":"fbed28f6322446dc8eadfbe009d22ea4948046f155fa7e2acd7d5cffafa63f9e","schema_version":"1.0","event_id":"sha256:fbed28f6322446dc8eadfbe009d22ea4948046f155fa7e2acd7d5cffafa63f9e"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/E3HL3C2TDH7CBKJ2KFCNA5LS6Y/bundle.json","state_url":"https://pith.science/pith/E3HL3C2TDH7CBKJ2KFCNA5LS6Y/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/E3HL3C2TDH7CBKJ2KFCNA5LS6Y/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":"2026-08-07T12:18:26Z","links":{"resolver":"https://pith.science/pith/E3HL3C2TDH7CBKJ2KFCNA5LS6Y","bundle":"https://pith.science/pith/E3HL3C2TDH7CBKJ2KFCNA5LS6Y/bundle.json","state":"https://pith.science/pith/E3HL3C2TDH7CBKJ2KFCNA5LS6Y/state.json","well_known_bundle":"https://pith.science/.well-known/pith/E3HL3C2TDH7CBKJ2KFCNA5LS6Y/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:E3HL3C2TDH7CBKJ2KFCNA5LS6Y","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":"13b1b26fb2523bc99d6cffa95c0dabef853cf9538037aefa96f87f41d6b9775c","cross_cats_sorted":["cs.LG","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-06T17:17:36Z","title_canon_sha256":"d44a0a341a6311250ee246d415ab9511beb21683367e7daf3a24c9f91b0bcdbd"},"schema_version":"1.0","source":{"id":"2403.03867","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.03867","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"arxiv_version","alias_value":"2403.03867v1","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.03867","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"pith_short_12","alias_value":"E3HL3C2TDH7C","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"pith_short_16","alias_value":"E3HL3C2TDH7CBKJ2","created_at":"2026-07-05T07:53:02Z"},{"alias_kind":"pith_short_8","alias_value":"E3HL3C2T","created_at":"2026-07-05T07:53:02Z"}],"graph_snapshots":[{"event_id":"sha256:fbed28f6322446dc8eadfbe009d22ea4948046f155fa7e2acd7d5cffafa63f9e","target":"graph","created_at":"2026-07-05T07:53:02Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2403.03867/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recent works have argued that high-level semantic concepts are encoded \"linearly\" in the representation space of large language models. In this work, we study the origins of such linear representations. To that end, we introduce a simple latent variable model to abstract and formalize the concept dynamics of the next token prediction. We use this formalism to show that the next token prediction objective (softmax with cross-entropy) and the implicit bias of gradient descent together promote the linear representation of concepts. Experiments show that linear representations emerge when learning","authors_text":"Bryon Aragam, Goutham Rajendran, Pradeep Ravikumar, Victor Veitch, Yibo Jiang","cross_cats":["cs.LG","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-06T17:17:36Z","title":"On the Origins of Linear Representations in Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.03867","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ec85e0b2fb7e131991e29d71dc3652caf0b5bf93ce95ca443b33b78d4003c039","target":"record","created_at":"2026-07-05T07:53:02Z","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":"13b1b26fb2523bc99d6cffa95c0dabef853cf9538037aefa96f87f41d6b9775c","cross_cats_sorted":["cs.LG","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-03-06T17:17:36Z","title_canon_sha256":"d44a0a341a6311250ee246d415ab9511beb21683367e7daf3a24c9f91b0bcdbd"},"schema_version":"1.0","source":{"id":"2403.03867","kind":"arxiv","version":1}},"canonical_sha256":"26cebd8b5319fe20a93a5144d07572f60856e0311cf091794bffb63cb1603877","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"26cebd8b5319fe20a93a5144d07572f60856e0311cf091794bffb63cb1603877","first_computed_at":"2026-07-05T07:53:02.499748Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:53:02.499748Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"qR923socxKAMZ1+FDsiWu1aPjp/KlIsWHAeXgJTNSD1WQVAI6Ckzl1CqYiijgW+2RSM1k+Jjp3q75T5A5S4sBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T07:53:02.500220Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.03867","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ec85e0b2fb7e131991e29d71dc3652caf0b5bf93ce95ca443b33b78d4003c039","sha256:fbed28f6322446dc8eadfbe009d22ea4948046f155fa7e2acd7d5cffafa63f9e"],"state_sha256":"7ca3b22ed21fece165e00894095b5e7a9226a9e86a655c5546025bb696ba6517"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"rdTqNXMHVneDaw33h1IuO8jWte31D/y7nfkomTtqJ0tLzaGhoXIBHcQBZulV/de5qTEWA/KI54OXUlG+iXPQAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T12:18:26.215479Z","bundle_sha256":"6fda23f58d6e85839771899d6a587f356c0df0ded2a38695bf4e57ad7ec90c78"}}