{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:TWIY36YBVYQFCU7JNKVAZNBYNB","short_pith_number":"pith:TWIY36YB","canonical_record":{"source":{"id":"2403.08081","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-12T21:15:38Z","cross_cats_sorted":["cs.AI","cs.CL","math.OC"],"title_canon_sha256":"bd59468fc7c2bfefc3cf7d060817ebed7206dd3d816458c394837c3c6af17f32","abstract_canon_sha256":"95976742ea4c2352ce9a42393734665873a891e1b8fe6a62cf24dcfa6d6a1877"},"schema_version":"1.0"},"canonical_sha256":"9d918dfb01ae205153e96aaa0cb4386846771ee1abb77a0c4aa480110de918dd","source":{"kind":"arxiv","id":"2403.08081","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.08081","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"arxiv_version","alias_value":"2403.08081v1","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.08081","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"pith_short_12","alias_value":"TWIY36YBVYQF","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"pith_short_16","alias_value":"TWIY36YBVYQFCU7J","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"pith_short_8","alias_value":"TWIY36YB","created_at":"2026-07-05T07:55:34Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:TWIY36YBVYQFCU7JNKVAZNBYNB","target":"record","payload":{"canonical_record":{"source":{"id":"2403.08081","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-12T21:15:38Z","cross_cats_sorted":["cs.AI","cs.CL","math.OC"],"title_canon_sha256":"bd59468fc7c2bfefc3cf7d060817ebed7206dd3d816458c394837c3c6af17f32","abstract_canon_sha256":"95976742ea4c2352ce9a42393734665873a891e1b8fe6a62cf24dcfa6d6a1877"},"schema_version":"1.0"},"canonical_sha256":"9d918dfb01ae205153e96aaa0cb4386846771ee1abb77a0c4aa480110de918dd","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:55:34.217170Z","signature_b64":"jg3OVKocQ4WoVrL98WrrZdRt0+1hRU7BeK3K9tYLMRr7tCSG3cq0odjvfGy1iyMuVyW8xPJYKYIXj/kEubyGDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9d918dfb01ae205153e96aaa0cb4386846771ee1abb77a0c4aa480110de918dd","last_reissued_at":"2026-07-05T07:55:34.216788Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:55:34.216788Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2403.08081","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":"2026-07-05T07:55:34Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Fz1YUrVsa35eBNFCVEG1WnF9MmlsPMtckTFIuZWqZhInrw8h126/vSLFl1Pobfh32XUcDyBNY9WeTByjV1BwBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T04:42:01.729880Z"},"content_sha256":"1d0b9b3e4a7850766907c28bfafc1d219a245a79125d83ee2ea12b289a599736","schema_version":"1.0","event_id":"sha256:1d0b9b3e4a7850766907c28bfafc1d219a245a79125d83ee2ea12b289a599736"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:TWIY36YBVYQFCU7JNKVAZNBYNB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Mechanics of Next Token Prediction with Self-Attention","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","math.OC"],"primary_cat":"cs.LG","authors_text":"Ankit Singh Rawat, M. Emrullah Ildiz, Samet Oymak, Yingcong Li, Yixiao Huang","submitted_at":"2024-03-12T21:15:38Z","abstract_excerpt":"Transformer-based language models are trained on large datasets to predict the next token given an input sequence. Despite this simple training objective, they have led to revolutionary advances in natural language processing. Underlying this success is the self-attention mechanism. In this work, we ask: $\\textit{What}$ $\\textit{does}$ $\\textit{a}$ $\\textit{single}$ $\\textit{self-attention}$ $\\textit{layer}$ $\\textit{learn}$ $\\textit{from}$ $\\textit{next-token}$ $\\textit{prediction?}$ We show that training self-attention with gradient descent learns an automaton which generates the next token "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.08081","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/2403.08081/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":"2026-07-05T07:55:34Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tFHxVCU1DVrMz9OLMdI8XFVWPVh4KwAK9OQfw3XMLTt/ay9+LcKl0DWwCEg597jWE5wSWhjnBm6u+dk7ZlM0Bw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T04:42:01.730792Z"},"content_sha256":"3e8c54cf48b1fa32259ca76ccb1e76db457489909062d22510e65b23d713e6fa","schema_version":"1.0","event_id":"sha256:3e8c54cf48b1fa32259ca76ccb1e76db457489909062d22510e65b23d713e6fa"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TWIY36YBVYQFCU7JNKVAZNBYNB/bundle.json","state_url":"https://pith.science/pith/TWIY36YBVYQFCU7JNKVAZNBYNB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TWIY36YBVYQFCU7JNKVAZNBYNB/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":"2026-08-17T04:42:01Z","links":{"resolver":"https://pith.science/pith/TWIY36YBVYQFCU7JNKVAZNBYNB","bundle":"https://pith.science/pith/TWIY36YBVYQFCU7JNKVAZNBYNB/bundle.json","state":"https://pith.science/pith/TWIY36YBVYQFCU7JNKVAZNBYNB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TWIY36YBVYQFCU7JNKVAZNBYNB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:TWIY36YBVYQFCU7JNKVAZNBYNB","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":"95976742ea4c2352ce9a42393734665873a891e1b8fe6a62cf24dcfa6d6a1877","cross_cats_sorted":["cs.AI","cs.CL","math.OC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-12T21:15:38Z","title_canon_sha256":"bd59468fc7c2bfefc3cf7d060817ebed7206dd3d816458c394837c3c6af17f32"},"schema_version":"1.0","source":{"id":"2403.08081","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.08081","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"arxiv_version","alias_value":"2403.08081v1","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.08081","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"pith_short_12","alias_value":"TWIY36YBVYQF","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"pith_short_16","alias_value":"TWIY36YBVYQFCU7J","created_at":"2026-07-05T07:55:34Z"},{"alias_kind":"pith_short_8","alias_value":"TWIY36YB","created_at":"2026-07-05T07:55:34Z"}],"graph_snapshots":[{"event_id":"sha256:3e8c54cf48b1fa32259ca76ccb1e76db457489909062d22510e65b23d713e6fa","target":"graph","created_at":"2026-07-05T07:55:34Z","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/2403.08081/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Transformer-based language models are trained on large datasets to predict the next token given an input sequence. Despite this simple training objective, they have led to revolutionary advances in natural language processing. Underlying this success is the self-attention mechanism. In this work, we ask: $\\textit{What}$ $\\textit{does}$ $\\textit{a}$ $\\textit{single}$ $\\textit{self-attention}$ $\\textit{layer}$ $\\textit{learn}$ $\\textit{from}$ $\\textit{next-token}$ $\\textit{prediction?}$ We show that training self-attention with gradient descent learns an automaton which generates the next token ","authors_text":"Ankit Singh Rawat, M. Emrullah Ildiz, Samet Oymak, Yingcong Li, Yixiao Huang","cross_cats":["cs.AI","cs.CL","math.OC"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-12T21:15:38Z","title":"Mechanics of Next Token Prediction with Self-Attention"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.08081","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":"sha256:1d0b9b3e4a7850766907c28bfafc1d219a245a79125d83ee2ea12b289a599736","target":"record","created_at":"2026-07-05T07:55:34Z","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":"95976742ea4c2352ce9a42393734665873a891e1b8fe6a62cf24dcfa6d6a1877","cross_cats_sorted":["cs.AI","cs.CL","math.OC"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-12T21:15:38Z","title_canon_sha256":"bd59468fc7c2bfefc3cf7d060817ebed7206dd3d816458c394837c3c6af17f32"},"schema_version":"1.0","source":{"id":"2403.08081","kind":"arxiv","version":1}},"canonical_sha256":"9d918dfb01ae205153e96aaa0cb4386846771ee1abb77a0c4aa480110de918dd","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9d918dfb01ae205153e96aaa0cb4386846771ee1abb77a0c4aa480110de918dd","first_computed_at":"2026-07-05T07:55:34.216788Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T07:55:34.216788Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"jg3OVKocQ4WoVrL98WrrZdRt0+1hRU7BeK3K9tYLMRr7tCSG3cq0odjvfGy1iyMuVyW8xPJYKYIXj/kEubyGDA==","signature_status":"signed_v1","signed_at":"2026-07-05T07:55:34.217170Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.08081","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1d0b9b3e4a7850766907c28bfafc1d219a245a79125d83ee2ea12b289a599736","sha256:3e8c54cf48b1fa32259ca76ccb1e76db457489909062d22510e65b23d713e6fa"],"state_sha256":"486d80053d4ab478e142e33acc1e032b555f066c51b741b8bcaffac75286d979"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1JmKmQ/jL7Ni6Okz7Eqi1RXyjx+/JV8uAIR237Jiuxr3v43KO3gakGaKEAQQpC7Ojbhg6/AXAn/xjZ9A+omzCg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T04:42:01.742784Z","bundle_sha256":"4aa2d39e39226c7652b4a73e22875bdd9565cbfcb0f74732f2e7f0f23a00c5fa"}}