{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:43A67ZYOJ5V4RHRI6OMMDGVJBW","short_pith_number":"pith:43A67ZYO","schema_version":"1.0","canonical_sha256":"e6c1efe70e4f6bc89e28f398c19aa90db6468aae810e82bd5c7a32c96f8291c5","source":{"kind":"arxiv","id":"2310.07820","version":3},"attestation_state":"computed","paper":{"title":"Large Language Models Are Zero-Shot Time Series Forecasters","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andrew Gordon Wilson, Marc Finzi, Nate Gruver, Shikai Qiu","submitted_at":"2023-10-11T19:01:28Z","abstract_excerpt":"By encoding time series as a string of numerical digits, we can frame time series forecasting as next-token prediction in text. Developing this approach, we find that large language models (LLMs) such as GPT-3 and LLaMA-2 can surprisingly zero-shot extrapolate time series at a level comparable to or exceeding the performance of purpose-built time series models trained on the downstream tasks. To facilitate this performance, we propose procedures for effectively tokenizing time series data and converting discrete distributions over tokens into highly flexible densities over continuous values. W"},"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":"2310.07820","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-10-11T19:01:28Z","cross_cats_sorted":[],"title_canon_sha256":"302ed2f8c7bb71da2ead629758897bc8cdb36fef3b33666dbbb1523013a429e0","abstract_canon_sha256":"7ac9cb655e725e903716318cd5f911fba42f42b2486fc0656aa349a350e9ff38"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:54:04.437507Z","signature_b64":"Rz4U3MGjHyPnFS07Cl/tcjgpZo6euxdmyclBzWrIQ8R+LSpYuBlg1R2eU7ay/vOK8MNIHRiQiHtCwjHezPATDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e6c1efe70e4f6bc89e28f398c19aa90db6468aae810e82bd5c7a32c96f8291c5","last_reissued_at":"2026-07-05T08:54:04.437023Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:54:04.437023Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large Language Models Are Zero-Shot Time Series Forecasters","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Andrew Gordon Wilson, Marc Finzi, Nate Gruver, Shikai Qiu","submitted_at":"2023-10-11T19:01:28Z","abstract_excerpt":"By encoding time series as a string of numerical digits, we can frame time series forecasting as next-token prediction in text. Developing this approach, we find that large language models (LLMs) such as GPT-3 and LLaMA-2 can surprisingly zero-shot extrapolate time series at a level comparable to or exceeding the performance of purpose-built time series models trained on the downstream tasks. To facilitate this performance, we propose procedures for effectively tokenizing time series data and converting discrete distributions over tokens into highly flexible densities over continuous values. W"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.07820","kind":"arxiv","version":3},"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/2310.07820/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":"2310.07820","created_at":"2026-07-05T08:54:04.437080+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.07820v3","created_at":"2026-07-05T08:54:04.437080+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.07820","created_at":"2026-07-05T08:54:04.437080+00:00"},{"alias_kind":"pith_short_12","alias_value":"43A67ZYOJ5V4","created_at":"2026-07-05T08:54:04.437080+00:00"},{"alias_kind":"pith_short_16","alias_value":"43A67ZYOJ5V4RHRI","created_at":"2026-07-05T08:54:04.437080+00:00"},{"alias_kind":"pith_short_8","alias_value":"43A67ZYO","created_at":"2026-07-05T08:54:04.437080+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06623","citing_title":"LLM-Guided Task-Semantic Field Factorization for Industrial Process Forecasting","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2606.11816","citing_title":"WorldReasoner: Evaluating Whether Language Model Agents Forecast Events with Valid Reasoning","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2606.10359","citing_title":"ReflectiChain: Epistemic Grounding in LLM-Driven World Models for Supply Chain Resilience","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05332","citing_title":"GITCO: Gated Inference-Time Context Optimization in TSFMs","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01634","citing_title":"E4GEN: Event-level Explainable Extreme-Enhanced Time-series Generation","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00384","citing_title":"VESTA: Visual Exploration with Statistical Tool Agents","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00241","citing_title":"InfoAtlas: A Foundation Model for Zero-Shot Statistical Dependence Estimate","ref_index":204,"is_internal_anchor":false},{"citing_arxiv_id":"2502.00816","citing_title":"Sundial: A Family of Highly Capable Time Series Foundation Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2506.07298","citing_title":"Pre-trained Large Language Models Learn Hidden Markov Models In-context","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2509.02967","citing_title":"AR-KAN: Autoregressive-Weight-Enhanced Kolmogorov-Arnold Network for Time Series Forecasting","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2511.01101","citing_title":"TSVer: A Benchmark for Fact Verification Against Time-Series Evidence","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2310.10688","citing_title":"A decoder-only foundation model for time-series forecasting","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14994","citing_title":"Degradation-aware Predictive Energy Management for Fuel Cell-Battery Ship Power System with Data-driven Load Forecasting","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW","json":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW.json","graph_json":"https://pith.science/api/pith-number/43A67ZYOJ5V4RHRI6OMMDGVJBW/graph.json","events_json":"https://pith.science/api/pith-number/43A67ZYOJ5V4RHRI6OMMDGVJBW/events.json","paper":"https://pith.science/paper/43A67ZYO"},"agent_actions":{"view_html":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW","download_json":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW.json","view_paper":"https://pith.science/paper/43A67ZYO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.07820&json=true","fetch_graph":"https://pith.science/api/pith-number/43A67ZYOJ5V4RHRI6OMMDGVJBW/graph.json","fetch_events":"https://pith.science/api/pith-number/43A67ZYOJ5V4RHRI6OMMDGVJBW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW/action/storage_attestation","attest_author":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW/action/author_attestation","sign_citation":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW/action/citation_signature","submit_replication":"https://pith.science/pith/43A67ZYOJ5V4RHRI6OMMDGVJBW/action/replication_record"}},"created_at":"2026-07-05T08:54:04.437080+00:00","updated_at":"2026-07-05T08:54:04.437080+00:00"}