{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:UDBRALD47AVPPO7VVCRZIMAJJ2","short_pith_number":"pith:UDBRALD4","schema_version":"1.0","canonical_sha256":"a0c3102c7cf82af7bbf5a8a39430094eab5c6b4f1f94524335f60d470f234bda","source":{"kind":"arxiv","id":"2312.16337","version":1},"attestation_state":"computed","paper":{"title":"Task Contamination: Language Models May Not Be Few-Shot Anymore","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Changmao Li, Jeffrey Flanigan","submitted_at":"2023-12-26T21:17:46Z","abstract_excerpt":"Large language models (LLMs) offer impressive performance in various zero-shot and few-shot tasks. However, their success in zero-shot and few-shot settings may be affected by task contamination, a potential limitation that has not been thoroughly examined. This paper investigates how zero-shot and few-shot performance of LLMs has changed chronologically over time. Utilizing GPT-3 series models and several other recent open-sourced LLMs, and controlling for dataset difficulty, we find that on datasets released before the LLM training data creation date, LLMs perform surprisingly better than on"},"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":"2312.16337","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-12-26T21:17:46Z","cross_cats_sorted":[],"title_canon_sha256":"a2a9916ddf7f159d995b3298d1b7a6bf4d2c1d841b3e819d01f74926804deafa","abstract_canon_sha256":"a75965fed69f63b4451365a177e028ffdf54841eee76acae04b5ff5e36ad1dcf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:28:59.342704Z","signature_b64":"ZqV/KHQxaeFJ7lCbNGxKLHNhP31cvVr3IxcZSxIfZxmKkkPYrpWq72UIvyMisgD9kQhePTNKTzNOi2X4X9N/AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a0c3102c7cf82af7bbf5a8a39430094eab5c6b4f1f94524335f60d470f234bda","last_reissued_at":"2026-07-05T07:28:59.342223Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:28:59.342223Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Task Contamination: Language Models May Not Be Few-Shot Anymore","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Changmao Li, Jeffrey Flanigan","submitted_at":"2023-12-26T21:17:46Z","abstract_excerpt":"Large language models (LLMs) offer impressive performance in various zero-shot and few-shot tasks. However, their success in zero-shot and few-shot settings may be affected by task contamination, a potential limitation that has not been thoroughly examined. This paper investigates how zero-shot and few-shot performance of LLMs has changed chronologically over time. Utilizing GPT-3 series models and several other recent open-sourced LLMs, and controlling for dataset difficulty, we find that on datasets released before the LLM training data creation date, LLMs perform surprisingly better than on"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.16337","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/2312.16337/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":"2312.16337","created_at":"2026-07-05T07:28:59.342284+00:00"},{"alias_kind":"arxiv_version","alias_value":"2312.16337v1","created_at":"2026-07-05T07:28:59.342284+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.16337","created_at":"2026-07-05T07:28:59.342284+00:00"},{"alias_kind":"pith_short_12","alias_value":"UDBRALD47AVP","created_at":"2026-07-05T07:28:59.342284+00:00"},{"alias_kind":"pith_short_16","alias_value":"UDBRALD47AVPPO7V","created_at":"2026-07-05T07:28:59.342284+00:00"},{"alias_kind":"pith_short_8","alias_value":"UDBRALD4","created_at":"2026-07-05T07:28:59.342284+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.26133","citing_title":"Pretraining Data Exposure in Large Language Models: A Survey of Membership Inference, Data Contamination, and Security Implications","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2403.07974","citing_title":"LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code","ref_index":208,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2","json":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2.json","graph_json":"https://pith.science/api/pith-number/UDBRALD47AVPPO7VVCRZIMAJJ2/graph.json","events_json":"https://pith.science/api/pith-number/UDBRALD47AVPPO7VVCRZIMAJJ2/events.json","paper":"https://pith.science/paper/UDBRALD4"},"agent_actions":{"view_html":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2","download_json":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2.json","view_paper":"https://pith.science/paper/UDBRALD4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2312.16337&json=true","fetch_graph":"https://pith.science/api/pith-number/UDBRALD47AVPPO7VVCRZIMAJJ2/graph.json","fetch_events":"https://pith.science/api/pith-number/UDBRALD47AVPPO7VVCRZIMAJJ2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2/action/storage_attestation","attest_author":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2/action/author_attestation","sign_citation":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2/action/citation_signature","submit_replication":"https://pith.science/pith/UDBRALD47AVPPO7VVCRZIMAJJ2/action/replication_record"}},"created_at":"2026-07-05T07:28:59.342284+00:00","updated_at":"2026-07-05T07:28:59.342284+00:00"}