{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MYPJ2XRD4KO4IHKTCCTSNJLTCV","short_pith_number":"pith:MYPJ2XRD","schema_version":"1.0","canonical_sha256":"661e9d5e23e29dc41d5310a726a573155e2dd740d72c2df6aafd78b70c84f536","source":{"kind":"arxiv","id":"2506.00723","version":1},"attestation_state":"computed","paper":{"title":"Pitfalls in Evaluating Language Model Forecasters","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR"],"primary_cat":"cs.LG","authors_text":"Daniel Paleka, Florian Tram\\`er, Jonas Geiping, Shashwat Goel","submitted_at":"2025-05-31T21:49:17Z","abstract_excerpt":"Large language models (LLMs) have recently been applied to forecasting tasks, with some works claiming these systems match or exceed human performance. In this paper, we argue that, as a community, we should be careful about such conclusions as evaluating LLM forecasters presents unique challenges. We identify two broad categories of issues: (1) difficulty in trusting evaluation results due to many forms of temporal leakage, and (2) difficulty in extrapolating from evaluation performance to real-world forecasting. Through systematic analysis and concrete examples from prior work, we demonstrat"},"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":"2506.00723","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-31T21:49:17Z","cross_cats_sorted":["cs.AI","cs.IR"],"title_canon_sha256":"86a65399116464c9bfc9be5edc183c391765db44113163997b9c57ee42b7647c","abstract_canon_sha256":"5a32d8c0c7fe159bf92f1b53d7ce54a5636b37ee112cc45d04d6dec78aba5376"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:13:39.333356Z","signature_b64":"+GP35eo3tHLQ6+wY3A+Vhna4TR6ofPzhj/i03L8U6Em940vYts0JSjD7iM7yOZUF5O7dfxNNvXZu5gZAYjwiAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"661e9d5e23e29dc41d5310a726a573155e2dd740d72c2df6aafd78b70c84f536","last_reissued_at":"2026-07-05T11:13:39.332781Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:13:39.332781Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pitfalls in Evaluating Language Model Forecasters","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR"],"primary_cat":"cs.LG","authors_text":"Daniel Paleka, Florian Tram\\`er, Jonas Geiping, Shashwat Goel","submitted_at":"2025-05-31T21:49:17Z","abstract_excerpt":"Large language models (LLMs) have recently been applied to forecasting tasks, with some works claiming these systems match or exceed human performance. In this paper, we argue that, as a community, we should be careful about such conclusions as evaluating LLM forecasters presents unique challenges. We identify two broad categories of issues: (1) difficulty in trusting evaluation results due to many forms of temporal leakage, and (2) difficulty in extrapolating from evaluation performance to real-world forecasting. Through systematic analysis and concrete examples from prior work, we demonstrat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.00723","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/2506.00723/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":"2506.00723","created_at":"2026-07-05T11:13:39.332845+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.00723v1","created_at":"2026-07-05T11:13:39.332845+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.00723","created_at":"2026-07-05T11:13:39.332845+00:00"},{"alias_kind":"pith_short_12","alias_value":"MYPJ2XRD4KO4","created_at":"2026-07-05T11:13:39.332845+00:00"},{"alias_kind":"pith_short_16","alias_value":"MYPJ2XRD4KO4IHKT","created_at":"2026-07-05T11:13:39.332845+00:00"},{"alias_kind":"pith_short_8","alias_value":"MYPJ2XRD","created_at":"2026-07-05T11:13:39.332845+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01661","citing_title":"Diverse Evidence, Better Forecasts: Multi-Agent Deliberation Under Information Asymmetry","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00164","citing_title":"Verifiable Rewards for Calibrated Probabilistic Forecasting","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22672","citing_title":"Is Capability a Liability? More Capable Language Models Make Worse Forecasts When It Matters Most","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22672","citing_title":"Is Capability a Liability? More Capable Language Models Make Worse Forecasts When It Matters Most","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03762","citing_title":"OracleProto: A Reproducible Framework for Benchmarking LLM Native Forecasting via Knowledge Cutoff and Temporal Masking","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18576","citing_title":"Agentic Forecasting using Sequential Bayesian Updating of Linguistic Beliefs","ref_index":33,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV","json":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV.json","graph_json":"https://pith.science/api/pith-number/MYPJ2XRD4KO4IHKTCCTSNJLTCV/graph.json","events_json":"https://pith.science/api/pith-number/MYPJ2XRD4KO4IHKTCCTSNJLTCV/events.json","paper":"https://pith.science/paper/MYPJ2XRD"},"agent_actions":{"view_html":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV","download_json":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV.json","view_paper":"https://pith.science/paper/MYPJ2XRD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.00723&json=true","fetch_graph":"https://pith.science/api/pith-number/MYPJ2XRD4KO4IHKTCCTSNJLTCV/graph.json","fetch_events":"https://pith.science/api/pith-number/MYPJ2XRD4KO4IHKTCCTSNJLTCV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV/action/storage_attestation","attest_author":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV/action/author_attestation","sign_citation":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV/action/citation_signature","submit_replication":"https://pith.science/pith/MYPJ2XRD4KO4IHKTCCTSNJLTCV/action/replication_record"}},"created_at":"2026-07-05T11:13:39.332845+00:00","updated_at":"2026-07-05T11:13:39.332845+00:00"}