{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:IUCB4GYJCEEL557S6G53ZUZ6VU","short_pith_number":"pith:IUCB4GYJ","canonical_record":{"source":{"id":"2406.03445","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-05T16:40:53Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"bde87c8f622650a94d987079bf7aca8fa9a7576fa3634da50b92c6ec06347011","abstract_canon_sha256":"8a9578c28a643c382d0a4116bef598001664fecfef1c66997c5f5a9d09034f36"},"schema_version":"1.0"},"canonical_sha256":"45041e1b091108bef7f2f1bbbcd33ead39f8612548a91c69d3add56199017bec","source":{"kind":"arxiv","id":"2406.03445","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.03445","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"arxiv_version","alias_value":"2406.03445v1","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.03445","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"pith_short_12","alias_value":"IUCB4GYJCEEL","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"pith_short_16","alias_value":"IUCB4GYJCEEL557S","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"pith_short_8","alias_value":"IUCB4GYJ","created_at":"2026-07-05T08:27:56Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:IUCB4GYJCEEL557S6G53ZUZ6VU","target":"record","payload":{"canonical_record":{"source":{"id":"2406.03445","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-05T16:40:53Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"bde87c8f622650a94d987079bf7aca8fa9a7576fa3634da50b92c6ec06347011","abstract_canon_sha256":"8a9578c28a643c382d0a4116bef598001664fecfef1c66997c5f5a9d09034f36"},"schema_version":"1.0"},"canonical_sha256":"45041e1b091108bef7f2f1bbbcd33ead39f8612548a91c69d3add56199017bec","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:27:56.825224Z","signature_b64":"8FGAJZBraZY0Dk0wS7tuy1xQIJJsTCIpYngsRxhnIMhdDJk3+teZ9EVanCybm9NCKJnJRUOXDAqVs/qS6XYUDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45041e1b091108bef7f2f1bbbcd33ead39f8612548a91c69d3add56199017bec","last_reissued_at":"2026-07-05T08:27:56.824755Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:27:56.824755Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.03445","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":"2026-07-05T08:27:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GdvP9e20kwZEtwYDmHV2AIX5LOsxoTGu7wIs7YawGW96WGrF90jkvsvy4xo/Au5hD1bCMrRrhuNSI/FLXBBoAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-13T18:23:39.164136Z"},"content_sha256":"5fae59f4111008dfef0c9ff059747e774eaf605bc7d97d098658c566e5c68578","schema_version":"1.0","event_id":"sha256:5fae59f4111008dfef0c9ff059747e774eaf605bc7d97d098658c566e5c68578"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:IUCB4GYJCEEL557S6G53ZUZ6VU","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Pre-trained Large Language Models Use Fourier Features to Compute Addition","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Deqing Fu, Robin Jia, Tianyi Zhou, Vatsal Sharan","submitted_at":"2024-06-05T16:40:53Z","abstract_excerpt":"Pre-trained large language models (LLMs) exhibit impressive mathematical reasoning capabilities, yet how they compute basic arithmetic, such as addition, remains unclear. This paper shows that pre-trained LLMs add numbers using Fourier features -- dimensions in the hidden state that represent numbers via a set of features sparse in the frequency domain. Within the model, MLP and attention layers use Fourier features in complementary ways: MLP layers primarily approximate the magnitude of the answer using low-frequency features, while attention layers primarily perform modular addition (e.g., c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.03445","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/2406.03445/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":"2026-07-05T08:27:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GZ2Gr7u+ayzERledaGwrxvljEki4jRTOV8R3BknlNY25P+0d9YH1PbXenkQ1Kz4TzcM4RvQsu0KK46nuMzamDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-13T18:23:39.164538Z"},"content_sha256":"e4a2dd94cfa4c444a344f7804ea277bb4334eada4924999a4561bdcf48483d8b","schema_version":"1.0","event_id":"sha256:e4a2dd94cfa4c444a344f7804ea277bb4334eada4924999a4561bdcf48483d8b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IUCB4GYJCEEL557S6G53ZUZ6VU/bundle.json","state_url":"https://pith.science/pith/IUCB4GYJCEEL557S6G53ZUZ6VU/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IUCB4GYJCEEL557S6G53ZUZ6VU/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":"2026-08-13T18:23:39Z","links":{"resolver":"https://pith.science/pith/IUCB4GYJCEEL557S6G53ZUZ6VU","bundle":"https://pith.science/pith/IUCB4GYJCEEL557S6G53ZUZ6VU/bundle.json","state":"https://pith.science/pith/IUCB4GYJCEEL557S6G53ZUZ6VU/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IUCB4GYJCEEL557S6G53ZUZ6VU/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:IUCB4GYJCEEL557S6G53ZUZ6VU","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":"8a9578c28a643c382d0a4116bef598001664fecfef1c66997c5f5a9d09034f36","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-05T16:40:53Z","title_canon_sha256":"bde87c8f622650a94d987079bf7aca8fa9a7576fa3634da50b92c6ec06347011"},"schema_version":"1.0","source":{"id":"2406.03445","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.03445","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"arxiv_version","alias_value":"2406.03445v1","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.03445","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"pith_short_12","alias_value":"IUCB4GYJCEEL","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"pith_short_16","alias_value":"IUCB4GYJCEEL557S","created_at":"2026-07-05T08:27:56Z"},{"alias_kind":"pith_short_8","alias_value":"IUCB4GYJ","created_at":"2026-07-05T08:27:56Z"}],"graph_snapshots":[{"event_id":"sha256:e4a2dd94cfa4c444a344f7804ea277bb4334eada4924999a4561bdcf48483d8b","target":"graph","created_at":"2026-07-05T08:27:56Z","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/2406.03445/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Pre-trained large language models (LLMs) exhibit impressive mathematical reasoning capabilities, yet how they compute basic arithmetic, such as addition, remains unclear. This paper shows that pre-trained LLMs add numbers using Fourier features -- dimensions in the hidden state that represent numbers via a set of features sparse in the frequency domain. Within the model, MLP and attention layers use Fourier features in complementary ways: MLP layers primarily approximate the magnitude of the answer using low-frequency features, while attention layers primarily perform modular addition (e.g., c","authors_text":"Deqing Fu, Robin Jia, Tianyi Zhou, Vatsal Sharan","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-05T16:40:53Z","title":"Pre-trained Large Language Models Use Fourier Features to Compute Addition"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.03445","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":"sha256:5fae59f4111008dfef0c9ff059747e774eaf605bc7d97d098658c566e5c68578","target":"record","created_at":"2026-07-05T08:27:56Z","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":"8a9578c28a643c382d0a4116bef598001664fecfef1c66997c5f5a9d09034f36","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-05T16:40:53Z","title_canon_sha256":"bde87c8f622650a94d987079bf7aca8fa9a7576fa3634da50b92c6ec06347011"},"schema_version":"1.0","source":{"id":"2406.03445","kind":"arxiv","version":1}},"canonical_sha256":"45041e1b091108bef7f2f1bbbcd33ead39f8612548a91c69d3add56199017bec","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"45041e1b091108bef7f2f1bbbcd33ead39f8612548a91c69d3add56199017bec","first_computed_at":"2026-07-05T08:27:56.824755Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:27:56.824755Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"8FGAJZBraZY0Dk0wS7tuy1xQIJJsTCIpYngsRxhnIMhdDJk3+teZ9EVanCybm9NCKJnJRUOXDAqVs/qS6XYUDg==","signature_status":"signed_v1","signed_at":"2026-07-05T08:27:56.825224Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.03445","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5fae59f4111008dfef0c9ff059747e774eaf605bc7d97d098658c566e5c68578","sha256:e4a2dd94cfa4c444a344f7804ea277bb4334eada4924999a4561bdcf48483d8b"],"state_sha256":"89fd1650e65645340536df140984c12eb58ee9af23bd594992b5afd0d5e4f974"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"X72IywXJShPLrDOoW4wK+EilNmKE4wKzCduWROKJgPrQkfHQc3svdrr/IPgYPMBMyovV4KNLfajwVc+ql0oUDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-13T18:23:39.167693Z","bundle_sha256":"b668000673b8b42a8798bb600da1ba6ed574a80b8ba1d896316cf75bb088bd7a"}}