{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:2MGJ2CYVSK2TS6VNH7JRI3IZQN","short_pith_number":"pith:2MGJ2CYV","schema_version":"1.0","canonical_sha256":"d30c9d0b1592b5397aad3fd3146d19837e988a73f51cf47164927a45415a51a8","source":{"kind":"arxiv","id":"2410.03011","version":2},"attestation_state":"computed","paper":{"title":"Towards Understanding the Universality of Transformers for Next-Token Prediction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Gabriel Peyr\\'e, Michael E. Sander","submitted_at":"2024-10-03T21:42:21Z","abstract_excerpt":"Causal Transformers are trained to predict the next token for a given context. While it is widely accepted that self-attention is crucial for encoding the causal structure of sequences, the precise underlying mechanism behind this in-context autoregressive learning ability remains unclear. In this paper, we take a step towards understanding this phenomenon by studying the approximation ability of Transformers for next-token prediction. Specifically, we explore the capacity of causal Transformers to predict the next token $x_{t+1}$ given an autoregressive sequence $(x_1, \\dots, x_t)$ as a promp"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2410.03011","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2024-10-03T21:42:21Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"7efa9d4476e522a0874b58652b72e584fb814dfbcfb117b21f3deeca882d4202","abstract_canon_sha256":"585ca118c89f5c74d1ae7c446a831eb35db97e7a9a2d32bc1aa3b03b0df12ab7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:22:05.696972Z","signature_b64":"xG6uXh5p2gixrTRp6TY3Xf8k8q2aYR5mo8fKhYcIUXJXxAbsayf+h2e/GmS/+WA+sRoaMlr4TGIAEbjEj/kSAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d30c9d0b1592b5397aad3fd3146d19837e988a73f51cf47164927a45415a51a8","last_reissued_at":"2026-07-05T10:22:05.696425Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:22:05.696425Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Understanding the Universality of Transformers for Next-Token Prediction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Gabriel Peyr\\'e, Michael E. Sander","submitted_at":"2024-10-03T21:42:21Z","abstract_excerpt":"Causal Transformers are trained to predict the next token for a given context. While it is widely accepted that self-attention is crucial for encoding the causal structure of sequences, the precise underlying mechanism behind this in-context autoregressive learning ability remains unclear. In this paper, we take a step towards understanding this phenomenon by studying the approximation ability of Transformers for next-token prediction. Specifically, we explore the capacity of causal Transformers to predict the next token $x_{t+1}$ given an autoregressive sequence $(x_1, \\dots, x_t)$ as a promp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.03011","kind":"arxiv","version":2},"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/2410.03011/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2410.03011","created_at":"2026-07-05T10:22:05.696489+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.03011v2","created_at":"2026-07-05T10:22:05.696489+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.03011","created_at":"2026-07-05T10:22:05.696489+00:00"},{"alias_kind":"pith_short_12","alias_value":"2MGJ2CYVSK2T","created_at":"2026-07-05T10:22:05.696489+00:00"},{"alias_kind":"pith_short_16","alias_value":"2MGJ2CYVSK2TS6VN","created_at":"2026-07-05T10:22:05.696489+00:00"},{"alias_kind":"pith_short_8","alias_value":"2MGJ2CYV","created_at":"2026-07-05T10:22:05.696489+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.28075","citing_title":"Measure-to-measure Regression with Transformers","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08475","citing_title":"Transformers Can Implement Preconditioned Richardson Iteration for In-Context Gaussian Kernel Regression","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08475","citing_title":"Transformers Can Implement Preconditioned Richardson Iteration for In-Context Gaussian Kernel Regression","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN","json":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN.json","graph_json":"https://pith.science/api/pith-number/2MGJ2CYVSK2TS6VNH7JRI3IZQN/graph.json","events_json":"https://pith.science/api/pith-number/2MGJ2CYVSK2TS6VNH7JRI3IZQN/events.json","paper":"https://pith.science/paper/2MGJ2CYV"},"agent_actions":{"view_html":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN","download_json":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN.json","view_paper":"https://pith.science/paper/2MGJ2CYV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.03011&json=true","fetch_graph":"https://pith.science/api/pith-number/2MGJ2CYVSK2TS6VNH7JRI3IZQN/graph.json","fetch_events":"https://pith.science/api/pith-number/2MGJ2CYVSK2TS6VNH7JRI3IZQN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN/action/storage_attestation","attest_author":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN/action/author_attestation","sign_citation":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN/action/citation_signature","submit_replication":"https://pith.science/pith/2MGJ2CYVSK2TS6VNH7JRI3IZQN/action/replication_record"}},"created_at":"2026-07-05T10:22:05.696489+00:00","updated_at":"2026-07-05T10:22:05.696489+00:00"}