{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:5YJMSOLNYWLYLBAGTNTNWFDZLW","short_pith_number":"pith:5YJMSOLN","schema_version":"1.0","canonical_sha256":"ee12c9396dc5978584069b66db14795db164c5040e8b67140955ed1a904af7f8","source":{"kind":"arxiv","id":"2202.00622","version":1},"attestation_state":"computed","paper":{"title":"Datamodels: Predicting Predictions from Training Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG"],"primary_cat":"stat.ML","authors_text":"Aleksander Madry, Andrew Ilyas, Guillaume Leclerc, Logan Engstrom, Sung Min Park","submitted_at":"2022-02-01T18:15:24Z","abstract_excerpt":"We present a conceptual framework, datamodeling, for analyzing the behavior of a model class in terms of the training data. For any fixed \"target\" example $x$, training set $S$, and learning algorithm, a datamodel is a parameterized function $2^S \\to \\mathbb{R}$ that for any subset of $S' \\subset S$ -- using only information about which examples of $S$ are contained in $S'$ -- predicts the outcome of training a model on $S'$ and evaluating on $x$. Despite the potential complexity of the underlying process being approximated (e.g., end-to-end training and evaluation of deep neural networks), we"},"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":"2202.00622","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2022-02-01T18:15:24Z","cross_cats_sorted":["cs.CV","cs.LG"],"title_canon_sha256":"c0014774ee99d4f026429145bf3591b0688656305c74252b3765a22cfbb25cfe","abstract_canon_sha256":"4ac00fb91b1664c6ef4ea9dd3adbfd241a7c6f9c916cea7961f11529fa7488b3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:53:23.831785Z","signature_b64":"IltvYzGlbkp4BjHJBEGTkB07uuYbA82HX5l7TI7cNmK2c4SoSv0IEZBTd0tm0wP6/vTkZ+zLiJHuSSrKZtBLBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ee12c9396dc5978584069b66db14795db164c5040e8b67140955ed1a904af7f8","last_reissued_at":"2026-07-05T03:53:23.831438Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:53:23.831438Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Datamodels: Predicting Predictions from Training Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.LG"],"primary_cat":"stat.ML","authors_text":"Aleksander Madry, Andrew Ilyas, Guillaume Leclerc, Logan Engstrom, Sung Min Park","submitted_at":"2022-02-01T18:15:24Z","abstract_excerpt":"We present a conceptual framework, datamodeling, for analyzing the behavior of a model class in terms of the training data. For any fixed \"target\" example $x$, training set $S$, and learning algorithm, a datamodel is a parameterized function $2^S \\to \\mathbb{R}$ that for any subset of $S' \\subset S$ -- using only information about which examples of $S$ are contained in $S'$ -- predicts the outcome of training a model on $S'$ and evaluating on $x$. Despite the potential complexity of the underlying process being approximated (e.g., end-to-end training and evaluation of deep neural networks), we"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2202.00622","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/2202.00622/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":"2202.00622","created_at":"2026-07-05T03:53:23.831495+00:00"},{"alias_kind":"arxiv_version","alias_value":"2202.00622v1","created_at":"2026-07-05T03:53:23.831495+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2202.00622","created_at":"2026-07-05T03:53:23.831495+00:00"},{"alias_kind":"pith_short_12","alias_value":"5YJMSOLNYWLY","created_at":"2026-07-05T03:53:23.831495+00:00"},{"alias_kind":"pith_short_16","alias_value":"5YJMSOLNYWLYLBAG","created_at":"2026-07-05T03:53:23.831495+00:00"},{"alias_kind":"pith_short_8","alias_value":"5YJMSOLN","created_at":"2026-07-05T03:53:23.831495+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00325","citing_title":"Watermarking for Proprietary Dataset Protection","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.24890","citing_title":"Small edits, large models: How Wikipedia advocacy shapes LLM values","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30462","citing_title":"idSCD: Identifying Training Datasets through Semantic Correlation Descriptors","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23591","citing_title":"Quantifying the Agreement Between Data-Influence and Data-Similarity to Understand LLM Behavior","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2310.12508","citing_title":"SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation","ref_index":94,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW","json":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW.json","graph_json":"https://pith.science/api/pith-number/5YJMSOLNYWLYLBAGTNTNWFDZLW/graph.json","events_json":"https://pith.science/api/pith-number/5YJMSOLNYWLYLBAGTNTNWFDZLW/events.json","paper":"https://pith.science/paper/5YJMSOLN"},"agent_actions":{"view_html":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW","download_json":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW.json","view_paper":"https://pith.science/paper/5YJMSOLN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2202.00622&json=true","fetch_graph":"https://pith.science/api/pith-number/5YJMSOLNYWLYLBAGTNTNWFDZLW/graph.json","fetch_events":"https://pith.science/api/pith-number/5YJMSOLNYWLYLBAGTNTNWFDZLW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW/action/storage_attestation","attest_author":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW/action/author_attestation","sign_citation":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW/action/citation_signature","submit_replication":"https://pith.science/pith/5YJMSOLNYWLYLBAGTNTNWFDZLW/action/replication_record"}},"created_at":"2026-07-05T03:53:23.831495+00:00","updated_at":"2026-07-05T03:53:23.831495+00:00"}