{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:I747WOMB52BGB3V3M7EL2GGLQM","short_pith_number":"pith:I747WOMB","schema_version":"1.0","canonical_sha256":"47f9fb3981ee8260eebb67c8bd18cb83374f6d99a7ac24fb076f7c9034b4d6a3","source":{"kind":"arxiv","id":"2606.24267","version":1},"attestation_state":"computed","paper":{"title":"Pigeonholing: Bad prompts hurt models to collapse and make mistakes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dorottya Demszky, Hyunji Nam, Keertana Chidambaram, Natasha Jaques","submitted_at":"2026-06-23T07:52:22Z","abstract_excerpt":"While in-context learning is generally shown to be effective in Large Language Models (LLMs), bad contexts can cause performance degradation and mode collapse, a phenomenon we call \"pigeonholing.\" **Unintentionally bad** contexts can happen without malicious jailbreaking intents: For example, a user asks the model to justify an incorrect math theorem or fails to correct the model's buggy code. Specifically, we investigate ``pigeonholing\" in two scenarios: (1) when the user suggests a solution, and (2) when the conversation context includes the assistant's previous (incorrect) responses. Our ex"},"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":"2606.24267","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-23T07:52:22Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c988261c3430b6ce5f688726209e902cd05726f168e8e21b02961f7c7cbb2fce","abstract_canon_sha256":"d5eb21fd6438c4102ee73c044b852162c68e76f8ce90271ed9d9f4b82410a308"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-24T01:14:48.525821Z","signature_b64":"VodNlbQYSND08PBO5zc2BBbJkrJdVui10KcvAgIXEQ2baaR46+iPAaWm/0i95+dapKaYGI8tGdgS2InP+rbmAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"47f9fb3981ee8260eebb67c8bd18cb83374f6d99a7ac24fb076f7c9034b4d6a3","last_reissued_at":"2026-06-24T01:14:48.525439Z","signature_status":"signed_v1","first_computed_at":"2026-06-24T01:14:48.525439Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pigeonholing: Bad prompts hurt models to collapse and make mistakes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dorottya Demszky, Hyunji Nam, Keertana Chidambaram, Natasha Jaques","submitted_at":"2026-06-23T07:52:22Z","abstract_excerpt":"While in-context learning is generally shown to be effective in Large Language Models (LLMs), bad contexts can cause performance degradation and mode collapse, a phenomenon we call \"pigeonholing.\" **Unintentionally bad** contexts can happen without malicious jailbreaking intents: For example, a user asks the model to justify an incorrect math theorem or fails to correct the model's buggy code. Specifically, we investigate ``pigeonholing\" in two scenarios: (1) when the user suggests a solution, and (2) when the conversation context includes the assistant's previous (incorrect) responses. Our ex"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.24267","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/2606.24267/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":"2606.24267","created_at":"2026-06-24T01:14:48.525493+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.24267v1","created_at":"2026-06-24T01:14:48.525493+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.24267","created_at":"2026-06-24T01:14:48.525493+00:00"},{"alias_kind":"pith_short_12","alias_value":"I747WOMB52BG","created_at":"2026-06-24T01:14:48.525493+00:00"},{"alias_kind":"pith_short_16","alias_value":"I747WOMB52BGB3V3","created_at":"2026-06-24T01:14:48.525493+00:00"},{"alias_kind":"pith_short_8","alias_value":"I747WOMB","created_at":"2026-06-24T01:14:48.525493+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM","json":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM.json","graph_json":"https://pith.science/api/pith-number/I747WOMB52BGB3V3M7EL2GGLQM/graph.json","events_json":"https://pith.science/api/pith-number/I747WOMB52BGB3V3M7EL2GGLQM/events.json","paper":"https://pith.science/paper/I747WOMB"},"agent_actions":{"view_html":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM","download_json":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM.json","view_paper":"https://pith.science/paper/I747WOMB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.24267&json=true","fetch_graph":"https://pith.science/api/pith-number/I747WOMB52BGB3V3M7EL2GGLQM/graph.json","fetch_events":"https://pith.science/api/pith-number/I747WOMB52BGB3V3M7EL2GGLQM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM/action/storage_attestation","attest_author":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM/action/author_attestation","sign_citation":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM/action/citation_signature","submit_replication":"https://pith.science/pith/I747WOMB52BGB3V3M7EL2GGLQM/action/replication_record"}},"created_at":"2026-06-24T01:14:48.525493+00:00","updated_at":"2026-06-24T01:14:48.525493+00:00"}