{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:CESDTY6K6CT7CDO3NEWIOX5HOO","short_pith_number":"pith:CESDTY6K","schema_version":"1.0","canonical_sha256":"112439e3caf0a7f10ddb692c875fa7738c56d2561e778036a1772f3fb488e427","source":{"kind":"arxiv","id":"2109.01247","version":2},"attestation_state":"computed","paper":{"title":"Do Prompt-Based Models Really Understand the Meaning of their Prompts?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Albert Webson, Ellie Pavlick","submitted_at":"2021-09-02T23:46:36Z","abstract_excerpt":"Recently, a boom of papers has shown extraordinary progress in zero-shot and few-shot learning with various prompt-based models. It is commonly argued that prompts help models to learn faster in the same way that humans learn faster when provided with task instructions expressed in natural language. In this study, we experiment with over 30 prompt templates manually written for natural language inference (NLI). We find that models learn just as fast with many prompts that are intentionally irrelevant or even pathologically misleading as they do with instructively \"good\" prompts. Further, such "},"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":"2109.01247","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-09-02T23:46:36Z","cross_cats_sorted":[],"title_canon_sha256":"c41701cfb24e48f07acffdde1d6922168269703f0e89bcc010f4c4ef32bf347f","abstract_canon_sha256":"cc0377ee47f1cd71bf5b5a10b61992ad298666d5b4df0e0af17d2a93f5b60544"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:16:36.426356Z","signature_b64":"wCi8UpYI8PmWvOsDhPrfkGrbsF96okU0FQoYfVHyerRj6No4zwtVoxX7bHP3qc+AyHVassPlE9XAV3Nw2flFBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"112439e3caf0a7f10ddb692c875fa7738c56d2561e778036a1772f3fb488e427","last_reissued_at":"2026-07-05T04:16:36.425889Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:16:36.425889Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Do Prompt-Based Models Really Understand the Meaning of their Prompts?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Albert Webson, Ellie Pavlick","submitted_at":"2021-09-02T23:46:36Z","abstract_excerpt":"Recently, a boom of papers has shown extraordinary progress in zero-shot and few-shot learning with various prompt-based models. It is commonly argued that prompts help models to learn faster in the same way that humans learn faster when provided with task instructions expressed in natural language. In this study, we experiment with over 30 prompt templates manually written for natural language inference (NLI). We find that models learn just as fast with many prompts that are intentionally irrelevant or even pathologically misleading as they do with instructively \"good\" prompts. Further, such "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.01247","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/2109.01247/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":"2109.01247","created_at":"2026-07-05T04:16:36.425943+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.01247v2","created_at":"2026-07-05T04:16:36.425943+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.01247","created_at":"2026-07-05T04:16:36.425943+00:00"},{"alias_kind":"pith_short_12","alias_value":"CESDTY6K6CT7","created_at":"2026-07-05T04:16:36.425943+00:00"},{"alias_kind":"pith_short_16","alias_value":"CESDTY6K6CT7CDO3","created_at":"2026-07-05T04:16:36.425943+00:00"},{"alias_kind":"pith_short_8","alias_value":"CESDTY6K","created_at":"2026-07-05T04:16:36.425943+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.04273","citing_title":"Characterizing initial human-AI proof formalization workflows","ref_index":214,"is_internal_anchor":false},{"citing_arxiv_id":"2211.01910","citing_title":"Large Language Models Are Human-Level Prompt Engineers","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2110.08207","citing_title":"Multitask Prompted Training Enables Zero-Shot Task Generalization","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2210.09261","citing_title":"Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Them","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO","json":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO.json","graph_json":"https://pith.science/api/pith-number/CESDTY6K6CT7CDO3NEWIOX5HOO/graph.json","events_json":"https://pith.science/api/pith-number/CESDTY6K6CT7CDO3NEWIOX5HOO/events.json","paper":"https://pith.science/paper/CESDTY6K"},"agent_actions":{"view_html":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO","download_json":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO.json","view_paper":"https://pith.science/paper/CESDTY6K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.01247&json=true","fetch_graph":"https://pith.science/api/pith-number/CESDTY6K6CT7CDO3NEWIOX5HOO/graph.json","fetch_events":"https://pith.science/api/pith-number/CESDTY6K6CT7CDO3NEWIOX5HOO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO/action/storage_attestation","attest_author":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO/action/author_attestation","sign_citation":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO/action/citation_signature","submit_replication":"https://pith.science/pith/CESDTY6K6CT7CDO3NEWIOX5HOO/action/replication_record"}},"created_at":"2026-07-05T04:16:36.425943+00:00","updated_at":"2026-07-05T04:16:36.425943+00:00"}