{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:NBRISUJR2GZZQRQGYJ23NLL3SI","short_pith_number":"pith:NBRISUJR","schema_version":"1.0","canonical_sha256":"6862895131d1b3984606c275b6ad7b923769ccac64029a13d6e546327af58c93","source":{"kind":"arxiv","id":"2212.10534","version":3},"attestation_state":"computed","paper":{"title":"DISCO: Distilling Counterfactuals with Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Antoine Bosselut, Ashish Sabharwal, Kyle Richardson, Qiyue Gao, Zeming Chen","submitted_at":"2022-12-20T18:46:08Z","abstract_excerpt":"Models trained with counterfactually augmented data learn representations of the causal structure of tasks, enabling robust generalization. However, high-quality counterfactual data is scarce for most tasks and not easily generated at scale. When crowdsourced, such data is typically limited in scale and diversity; when generated using supervised methods, it is computationally expensive to extend to new counterfactual dimensions. In this work, we introduce DISCO (DIStilled COunterfactual Data), a new method for automatically generating high quality counterfactual data at scale. DISCO engineers "},"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":"2212.10534","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-12-20T18:46:08Z","cross_cats_sorted":[],"title_canon_sha256":"dc4220d1b55a3df30b1718ba46fde8e9ae6708d6b87a9acd5370ebcffba0d703","abstract_canon_sha256":"558d09c1a2289e1c2c0b4d6556b698b913b2c8c52077a00ecee655ff196e9823"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:17:43.671187Z","signature_b64":"J0cmeRGuLHov2nItIXUVLOlmXicbPWUhRTLqli4fDGtb/Bsx7bbGhu8BUv17b4IPZ1Hc8Xq8VI1sX1cAJaJXBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6862895131d1b3984606c275b6ad7b923769ccac64029a13d6e546327af58c93","last_reissued_at":"2026-07-05T06:17:43.670713Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:17:43.670713Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DISCO: Distilling Counterfactuals with Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Antoine Bosselut, Ashish Sabharwal, Kyle Richardson, Qiyue Gao, Zeming Chen","submitted_at":"2022-12-20T18:46:08Z","abstract_excerpt":"Models trained with counterfactually augmented data learn representations of the causal structure of tasks, enabling robust generalization. However, high-quality counterfactual data is scarce for most tasks and not easily generated at scale. When crowdsourced, such data is typically limited in scale and diversity; when generated using supervised methods, it is computationally expensive to extend to new counterfactual dimensions. In this work, we introduce DISCO (DIStilled COunterfactual Data), a new method for automatically generating high quality counterfactual data at scale. DISCO engineers "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.10534","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/2212.10534/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":"2212.10534","created_at":"2026-07-05T06:17:43.670770+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.10534v3","created_at":"2026-07-05T06:17:43.670770+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.10534","created_at":"2026-07-05T06:17:43.670770+00:00"},{"alias_kind":"pith_short_12","alias_value":"NBRISUJR2GZZ","created_at":"2026-07-05T06:17:43.670770+00:00"},{"alias_kind":"pith_short_16","alias_value":"NBRISUJR2GZZQRQG","created_at":"2026-07-05T06:17:43.670770+00:00"},{"alias_kind":"pith_short_8","alias_value":"NBRISUJR","created_at":"2026-07-05T06:17:43.670770+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.25977","citing_title":"NeuronMLP: Efficient LLM Inference via Singular Value Decomposition Compression and Tiling on AWS Trainium","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI","json":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI.json","graph_json":"https://pith.science/api/pith-number/NBRISUJR2GZZQRQGYJ23NLL3SI/graph.json","events_json":"https://pith.science/api/pith-number/NBRISUJR2GZZQRQGYJ23NLL3SI/events.json","paper":"https://pith.science/paper/NBRISUJR"},"agent_actions":{"view_html":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI","download_json":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI.json","view_paper":"https://pith.science/paper/NBRISUJR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.10534&json=true","fetch_graph":"https://pith.science/api/pith-number/NBRISUJR2GZZQRQGYJ23NLL3SI/graph.json","fetch_events":"https://pith.science/api/pith-number/NBRISUJR2GZZQRQGYJ23NLL3SI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI/action/storage_attestation","attest_author":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI/action/author_attestation","sign_citation":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI/action/citation_signature","submit_replication":"https://pith.science/pith/NBRISUJR2GZZQRQGYJ23NLL3SI/action/replication_record"}},"created_at":"2026-07-05T06:17:43.670770+00:00","updated_at":"2026-07-05T06:17:43.670770+00:00"}