{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:E7FQG7PC6B63OEY3DTYXDCLGDZ","short_pith_number":"pith:E7FQG7PC","canonical_record":{"source":{"id":"2605.12924","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T03:00:08Z","cross_cats_sorted":[],"title_canon_sha256":"bc030469ff6725ab1fd0853ba1dc7ebffc386d8c5301c47cc181f318e40720a3","abstract_canon_sha256":"325cb2fb8dd13b92ae7c371449d3d7645808a2d2ea13c6b147ae254c4d6178de"},"schema_version":"1.0"},"canonical_sha256":"27cb037de2f07db7131b1cf17189661e68809cd2f1908e1b563a85d91ccf5c15","source":{"kind":"arxiv","id":"2605.12924","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12924","created_at":"2026-05-18T03:09:10Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12924v1","created_at":"2026-05-18T03:09:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12924","created_at":"2026-05-18T03:09:10Z"},{"alias_kind":"pith_short_12","alias_value":"E7FQG7PC6B63","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"E7FQG7PC6B63OEY3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"E7FQG7PC","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:E7FQG7PC6B63OEY3DTYXDCLGDZ","target":"record","payload":{"canonical_record":{"source":{"id":"2605.12924","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T03:00:08Z","cross_cats_sorted":[],"title_canon_sha256":"bc030469ff6725ab1fd0853ba1dc7ebffc386d8c5301c47cc181f318e40720a3","abstract_canon_sha256":"325cb2fb8dd13b92ae7c371449d3d7645808a2d2ea13c6b147ae254c4d6178de"},"schema_version":"1.0"},"canonical_sha256":"27cb037de2f07db7131b1cf17189661e68809cd2f1908e1b563a85d91ccf5c15","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:09:10.136737Z","signature_b64":"dXzPLI+HXUJkbUC+7kb3kXt/r18bSJk0RKwzOrHv5VB/dDLlytLqf21yWuXT0nqZdTH4qO4Hd5FsKRlPOE0jCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"27cb037de2f07db7131b1cf17189661e68809cd2f1908e1b563a85d91ccf5c15","last_reissued_at":"2026-05-18T03:09:10.136109Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:09:10.136109Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.12924","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:09:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"H2zVfeP9dgdABTFvUiU7Zl3Jjop7zQG2TXdf8ijT4Z857wcSuesliAxKqJP1FQ4uXfYdRXUBiQnH1tVKcgbuBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-21T19:31:44.706948Z"},"content_sha256":"d2ac9b4a5ab53ae406037fbdf2434d8788fc305a4df58d09ccf7dc0ee09b6f2b","schema_version":"1.0","event_id":"sha256:d2ac9b4a5ab53ae406037fbdf2434d8788fc305a4df58d09ccf7dc0ee09b6f2b"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:E7FQG7PC6B63OEY3DTYXDCLGDZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"IV-ICL: Bounding Causal Effects with Instrumental Variables via In-Context Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"An amortized in-context learner recovers the full identified set of causal effects from instrumental variable data.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hamidreza Kamkari, Medha Barath, Rahul G. Krishnan, Ricardo Silva, Vahid Balazadeh","submitted_at":"2026-05-13T03:00:08Z","abstract_excerpt":"The instrumental-variables (IV) setting is standard for partial identification of causal effects when unobserved confounding makes point identification impossible. Existing approaches face methodological bottlenecks: closed-form bound estimands are required -- e.g., Balke-Pearl equations in binary IV -- and even when available, designing accurate estimators requires manual effort tailored to each estimand. While direct Bayesian inference of the causal effects, instead of the bounds, circumvents these challenges, it is often computationally intensive and suffers from high prior sensitivity or u"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We introduce IV-ICL, an amortized Bayesian in-context learning method that learns the marginal posterior distribution of the causal effects directly and derives bounds as its quantiles. ... optimizing inclusive KL can recover the entire identified set across diverse data-generating processes, while exclusive-KL ... collapses onto a single mode.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the inclusive-KL objective in the amortized in-context learner will reliably recover the full identified set for arbitrary data-generating processes rather than only for the synthetic distributions used in training.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"IV-ICL learns the marginal posterior of causal effects via in-context learning to derive bounds as quantiles, recovering the identified set more reliably than variational inference while running 20-500x faster.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"An amortized in-context learner recovers the full identified set of causal effects from instrumental variable data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f6ac9d59d30c5724d4017fe97d3c105e5f251927c1914e489a364131294a3334"},"source":{"id":"2605.12924","kind":"arxiv","version":1},"verdict":{"id":"f76d2b26-ee45-4132-bfe9-9fd2db3d1101","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:45:13.789703Z","strongest_claim":"We introduce IV-ICL, an amortized Bayesian in-context learning method that learns the marginal posterior distribution of the causal effects directly and derives bounds as its quantiles. ... optimizing inclusive KL can recover the entire identified set across diverse data-generating processes, while exclusive-KL ... collapses onto a single mode.","one_line_summary":"IV-ICL learns the marginal posterior of causal effects via in-context learning to derive bounds as quantiles, recovering the identified set more reliably than variational inference while running 20-500x faster.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the inclusive-KL objective in the amortized in-context learner will reliably recover the full identified set for arbitrary data-generating processes rather than only for the synthetic distributions used in training.","pith_extraction_headline":"An amortized in-context learner recovers the full identified set of causal effects from instrumental variable data."},"references":{"count":75,"sample":[{"doi":"","year":2011,"title":"Accountability and flexibility in public schools: Evidence from boston’s charters and pilots.The Quarterly Journal of Economics, 126(2):699–748, 2011","work_id":"ad73b14e-a2cb-4cb8-895f-f476b0d97dda","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1994,"title":"Joshua D Angrist and Guido W. Imbens. Identification and estimation of local average treatment effects.Econometrica, 62:467–475, 1994","work_id":"a2abedde-25d7-4330-a77e-773f67d68c34","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2009,"title":"Princeton university press, 2009","work_id":"43435707-665d-4bf0-936f-d37c453cfb82","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1996,"title":"Identification of causal effects using instrumental variables.Journal of the American statistical Association, 91(434):444–455, 1996","work_id":"ebb3892f-7935-4e2c-a021-58ef8d61a3d7","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1994,"title":"The paired availability design: a proposal for evaluating epidural analgesia during labor.Statistics in medicine, 13(21):2269–2278, 1994","work_id":"0a7dc6c9-95f5-4213-9038-5a7a34096206","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":75,"snapshot_sha256":"b3baf0fea1520141a66ec905bbdd73411fbf0ed76a465e31ce77d54b943a91bc","internal_anchors":1},"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"},"verdict_id":"f76d2b26-ee45-4132-bfe9-9fd2db3d1101"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:09:10Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BrrRb1NYeKg3OMhnY3FwEHzxi/L3AoBUpsbxOfXrNl0FAnxJyS2RzWEPMijVm94jGBCDyObcABr3QCPCEvJiDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-21T19:31:44.708130Z"},"content_sha256":"d6c6bc9066a35ab597c95b301da7f05fc1bc8a3ec54e0f6bf728dd5eefa37ec1","schema_version":"1.0","event_id":"sha256:d6c6bc9066a35ab597c95b301da7f05fc1bc8a3ec54e0f6bf728dd5eefa37ec1"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/E7FQG7PC6B63OEY3DTYXDCLGDZ/bundle.json","state_url":"https://pith.science/pith/E7FQG7PC6B63OEY3DTYXDCLGDZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/E7FQG7PC6B63OEY3DTYXDCLGDZ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-21T19:31:44Z","links":{"resolver":"https://pith.science/pith/E7FQG7PC6B63OEY3DTYXDCLGDZ","bundle":"https://pith.science/pith/E7FQG7PC6B63OEY3DTYXDCLGDZ/bundle.json","state":"https://pith.science/pith/E7FQG7PC6B63OEY3DTYXDCLGDZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/E7FQG7PC6B63OEY3DTYXDCLGDZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:E7FQG7PC6B63OEY3DTYXDCLGDZ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"325cb2fb8dd13b92ae7c371449d3d7645808a2d2ea13c6b147ae254c4d6178de","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T03:00:08Z","title_canon_sha256":"bc030469ff6725ab1fd0853ba1dc7ebffc386d8c5301c47cc181f318e40720a3"},"schema_version":"1.0","source":{"id":"2605.12924","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12924","created_at":"2026-05-18T03:09:10Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12924v1","created_at":"2026-05-18T03:09:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12924","created_at":"2026-05-18T03:09:10Z"},{"alias_kind":"pith_short_12","alias_value":"E7FQG7PC6B63","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"E7FQG7PC6B63OEY3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"E7FQG7PC","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:d6c6bc9066a35ab597c95b301da7f05fc1bc8a3ec54e0f6bf728dd5eefa37ec1","target":"graph","created_at":"2026-05-18T03:09:10Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We introduce IV-ICL, an amortized Bayesian in-context learning method that learns the marginal posterior distribution of the causal effects directly and derives bounds as its quantiles. ... optimizing inclusive KL can recover the entire identified set across diverse data-generating processes, while exclusive-KL ... collapses onto a single mode."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the inclusive-KL objective in the amortized in-context learner will reliably recover the full identified set for arbitrary data-generating processes rather than only for the synthetic distributions used in training."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"IV-ICL learns the marginal posterior of causal effects via in-context learning to derive bounds as quantiles, recovering the identified set more reliably than variational inference while running 20-500x faster."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"An amortized in-context learner recovers the full identified set of causal effects from instrumental variable data."}],"snapshot_sha256":"f6ac9d59d30c5724d4017fe97d3c105e5f251927c1914e489a364131294a3334"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"The instrumental-variables (IV) setting is standard for partial identification of causal effects when unobserved confounding makes point identification impossible. Existing approaches face methodological bottlenecks: closed-form bound estimands are required -- e.g., Balke-Pearl equations in binary IV -- and even when available, designing accurate estimators requires manual effort tailored to each estimand. While direct Bayesian inference of the causal effects, instead of the bounds, circumvents these challenges, it is often computationally intensive and suffers from high prior sensitivity or u","authors_text":"Hamidreza Kamkari, Medha Barath, Rahul G. Krishnan, Ricardo Silva, Vahid Balazadeh","cross_cats":[],"headline":"An amortized in-context learner recovers the full identified set of causal effects from instrumental variable data.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T03:00:08Z","title":"IV-ICL: Bounding Causal Effects with Instrumental Variables via In-Context Learning"},"references":{"count":75,"internal_anchors":1,"resolved_work":75,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Accountability and flexibility in public schools: Evidence from boston’s charters and pilots.The Quarterly Journal of Economics, 126(2):699–748, 2011","work_id":"ad73b14e-a2cb-4cb8-895f-f476b0d97dda","year":2011},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Joshua D Angrist and Guido W. Imbens. Identification and estimation of local average treatment effects.Econometrica, 62:467–475, 1994","work_id":"a2abedde-25d7-4330-a77e-773f67d68c34","year":1994},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Princeton university press, 2009","work_id":"43435707-665d-4bf0-936f-d37c453cfb82","year":2009},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Identification of causal effects using instrumental variables.Journal of the American statistical Association, 91(434):444–455, 1996","work_id":"ebb3892f-7935-4e2c-a021-58ef8d61a3d7","year":1996},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"The paired availability design: a proposal for evaluating epidural analgesia during labor.Statistics in medicine, 13(21):2269–2278, 1994","work_id":"0a7dc6c9-95f5-4213-9038-5a7a34096206","year":1994}],"snapshot_sha256":"b3baf0fea1520141a66ec905bbdd73411fbf0ed76a465e31ce77d54b943a91bc"},"source":{"id":"2605.12924","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T19:45:13.789703Z","id":"f76d2b26-ee45-4132-bfe9-9fd2db3d1101","model_set":{"reader":"grok-4.3"},"one_line_summary":"IV-ICL learns the marginal posterior of causal effects via in-context learning to derive bounds as quantiles, recovering the identified set more reliably than variational inference while running 20-500x faster.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"An amortized in-context learner recovers the full identified set of causal effects from instrumental variable data.","strongest_claim":"We introduce IV-ICL, an amortized Bayesian in-context learning method that learns the marginal posterior distribution of the causal effects directly and derives bounds as its quantiles. ... optimizing inclusive KL can recover the entire identified set across diverse data-generating processes, while exclusive-KL ... collapses onto a single mode.","weakest_assumption":"That the inclusive-KL objective in the amortized in-context learner will reliably recover the full identified set for arbitrary data-generating processes rather than only for the synthetic distributions used in training."}},"verdict_id":"f76d2b26-ee45-4132-bfe9-9fd2db3d1101"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:d2ac9b4a5ab53ae406037fbdf2434d8788fc305a4df58d09ccf7dc0ee09b6f2b","target":"record","created_at":"2026-05-18T03:09:10Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"325cb2fb8dd13b92ae7c371449d3d7645808a2d2ea13c6b147ae254c4d6178de","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T03:00:08Z","title_canon_sha256":"bc030469ff6725ab1fd0853ba1dc7ebffc386d8c5301c47cc181f318e40720a3"},"schema_version":"1.0","source":{"id":"2605.12924","kind":"arxiv","version":1}},"canonical_sha256":"27cb037de2f07db7131b1cf17189661e68809cd2f1908e1b563a85d91ccf5c15","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"27cb037de2f07db7131b1cf17189661e68809cd2f1908e1b563a85d91ccf5c15","first_computed_at":"2026-05-18T03:09:10.136109Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:10.136109Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"dXzPLI+HXUJkbUC+7kb3kXt/r18bSJk0RKwzOrHv5VB/dDLlytLqf21yWuXT0nqZdTH4qO4Hd5FsKRlPOE0jCg==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:10.136737Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12924","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:d2ac9b4a5ab53ae406037fbdf2434d8788fc305a4df58d09ccf7dc0ee09b6f2b","sha256:d6c6bc9066a35ab597c95b301da7f05fc1bc8a3ec54e0f6bf728dd5eefa37ec1"],"state_sha256":"44b14746d82a7c7357c03e1504ab76ba691816176834ef3129f6ce57de41e42e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"etcempJTCn52cJLZ1xQhggIoPnXS7njXg3EJ8JMYkX0mAzaWV2/ilMtBFTG/ERq3eanVdD2ZLAosnAvGdBXkAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-21T19:31:44.712130Z","bundle_sha256":"c98917253415721a3f8c45d6e852cb579df89709114819b745f1e4d9e1a7cc81"}}