{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2021:ZYSCY7WOHTMWK7F7Y5JCD4HDV3","short_pith_number":"pith:ZYSCY7WO","canonical_record":{"source":{"id":"2102.02515","kind":"arxiv","version":7},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-04T10:00:13Z","cross_cats_sorted":[],"title_canon_sha256":"dc1d41d49fb07b2ebb99cba36102c43f0eebe9f1848772dc655e0807390e3c73","abstract_canon_sha256":"4d938a3ac643a7490e7ffcce9d28a6766fa9069f5eefd63af9a9fca32c489d64"},"schema_version":"1.0"},"canonical_sha256":"ce242c7ece3cd9657cbfc75221f0e3aed7121084525e359cda11b97709611835","source":{"kind":"arxiv","id":"2102.02515","version":7},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2102.02515","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"arxiv_version","alias_value":"2102.02515v7","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.02515","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"pith_short_12","alias_value":"ZYSCY7WOHTMW","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"pith_short_16","alias_value":"ZYSCY7WOHTMWK7F7","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"pith_short_8","alias_value":"ZYSCY7WO","created_at":"2026-07-05T10:54:28Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2021:ZYSCY7WOHTMWK7F7Y5JCD4HDV3","target":"record","payload":{"canonical_record":{"source":{"id":"2102.02515","kind":"arxiv","version":7},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-04T10:00:13Z","cross_cats_sorted":[],"title_canon_sha256":"dc1d41d49fb07b2ebb99cba36102c43f0eebe9f1848772dc655e0807390e3c73","abstract_canon_sha256":"4d938a3ac643a7490e7ffcce9d28a6766fa9069f5eefd63af9a9fca32c489d64"},"schema_version":"1.0"},"canonical_sha256":"ce242c7ece3cd9657cbfc75221f0e3aed7121084525e359cda11b97709611835","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:54:28.498473Z","signature_b64":"ACB2DJOnZX9e0K58vmSnpFhLk2BnL1TZ8kFTHh6rf4+fG4c6iyNp2rb96ZB3Bo/MMbhjj+IreW3S8J8mBTYWCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ce242c7ece3cd9657cbfc75221f0e3aed7121084525e359cda11b97709611835","last_reissued_at":"2026-07-05T10:54:28.497964Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:54:28.497964Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2102.02515","source_version":7,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:54:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"i+ubexZf0gUrz4gIYFRCb+mdwIjxTt73FaopJRyc5lJd3dV+vh83+1tkH8tbDAReSrAKTCKfOyop3cGgqVWwAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T22:52:48.382331Z"},"content_sha256":"0f5f328258738c36203f1026234c15f5070c302014e2d152b4659e5f4f91b52d","schema_version":"1.0","event_id":"sha256:0f5f328258738c36203f1026234c15f5070c302014e2d152b4659e5f4f91b52d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2021:ZYSCY7WOHTMWK7F7Y5JCD4HDV3","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"HYDRA: Hypergradient Data Relevance Analysis for Interpreting Deep Neural Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Boyang Li, Chunyan Miao, Han Yu, Pengcheng Wu, Yuanyuan Chen","submitted_at":"2021-02-04T10:00:13Z","abstract_excerpt":"The behaviors of deep neural networks (DNNs) are notoriously resistant to human interpretations. In this paper, we propose Hypergradient Data Relevance Analysis, or HYDRA, which interprets the predictions made by DNNs as effects of their training data. Existing approaches generally estimate data contributions around the final model parameters and ignore how the training data shape the optimization trajectory. By unrolling the hypergradient of test loss w.r.t. the weights of training data, HYDRA assesses the contribution of training data toward test data points throughout the training trajector"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.02515","kind":"arxiv","version":7},"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/2102.02515/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:54:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"3B6yKXfYLoBP9aH7cdiR6yo3+SWtxVKdW/+WoJwO8mciBcg+75pFFT0BtNUIJ80qFDZ5N/IwhqsHva4eu6t5BA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-03T22:52:48.382878Z"},"content_sha256":"11fa298f11eae4dca88ea9533d137b46351a42f22bedf562a0a3a4fd722ea0cb","schema_version":"1.0","event_id":"sha256:11fa298f11eae4dca88ea9533d137b46351a42f22bedf562a0a3a4fd722ea0cb"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/ZYSCY7WOHTMWK7F7Y5JCD4HDV3/bundle.json","state_url":"https://pith.science/pith/ZYSCY7WOHTMWK7F7Y5JCD4HDV3/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/ZYSCY7WOHTMWK7F7Y5JCD4HDV3/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":"2026-08-03T22:52:48Z","links":{"resolver":"https://pith.science/pith/ZYSCY7WOHTMWK7F7Y5JCD4HDV3","bundle":"https://pith.science/pith/ZYSCY7WOHTMWK7F7Y5JCD4HDV3/bundle.json","state":"https://pith.science/pith/ZYSCY7WOHTMWK7F7Y5JCD4HDV3/state.json","well_known_bundle":"https://pith.science/.well-known/pith/ZYSCY7WOHTMWK7F7Y5JCD4HDV3/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2021:ZYSCY7WOHTMWK7F7Y5JCD4HDV3","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":"4d938a3ac643a7490e7ffcce9d28a6766fa9069f5eefd63af9a9fca32c489d64","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-04T10:00:13Z","title_canon_sha256":"dc1d41d49fb07b2ebb99cba36102c43f0eebe9f1848772dc655e0807390e3c73"},"schema_version":"1.0","source":{"id":"2102.02515","kind":"arxiv","version":7}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2102.02515","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"arxiv_version","alias_value":"2102.02515v7","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.02515","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"pith_short_12","alias_value":"ZYSCY7WOHTMW","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"pith_short_16","alias_value":"ZYSCY7WOHTMWK7F7","created_at":"2026-07-05T10:54:28Z"},{"alias_kind":"pith_short_8","alias_value":"ZYSCY7WO","created_at":"2026-07-05T10:54:28Z"}],"graph_snapshots":[{"event_id":"sha256:11fa298f11eae4dca88ea9533d137b46351a42f22bedf562a0a3a4fd722ea0cb","target":"graph","created_at":"2026-07-05T10:54:28Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2102.02515/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The behaviors of deep neural networks (DNNs) are notoriously resistant to human interpretations. In this paper, we propose Hypergradient Data Relevance Analysis, or HYDRA, which interprets the predictions made by DNNs as effects of their training data. Existing approaches generally estimate data contributions around the final model parameters and ignore how the training data shape the optimization trajectory. By unrolling the hypergradient of test loss w.r.t. the weights of training data, HYDRA assesses the contribution of training data toward test data points throughout the training trajector","authors_text":"Boyang Li, Chunyan Miao, Han Yu, Pengcheng Wu, Yuanyuan Chen","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-04T10:00:13Z","title":"HYDRA: Hypergradient Data Relevance Analysis for Interpreting Deep Neural Networks"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.02515","kind":"arxiv","version":7},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:0f5f328258738c36203f1026234c15f5070c302014e2d152b4659e5f4f91b52d","target":"record","created_at":"2026-07-05T10:54:28Z","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":"4d938a3ac643a7490e7ffcce9d28a6766fa9069f5eefd63af9a9fca32c489d64","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-02-04T10:00:13Z","title_canon_sha256":"dc1d41d49fb07b2ebb99cba36102c43f0eebe9f1848772dc655e0807390e3c73"},"schema_version":"1.0","source":{"id":"2102.02515","kind":"arxiv","version":7}},"canonical_sha256":"ce242c7ece3cd9657cbfc75221f0e3aed7121084525e359cda11b97709611835","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ce242c7ece3cd9657cbfc75221f0e3aed7121084525e359cda11b97709611835","first_computed_at":"2026-07-05T10:54:28.497964Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:54:28.497964Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ACB2DJOnZX9e0K58vmSnpFhLk2BnL1TZ8kFTHh6rf4+fG4c6iyNp2rb96ZB3Bo/MMbhjj+IreW3S8J8mBTYWCA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:54:28.498473Z","signed_message":"canonical_sha256_bytes"},"source_id":"2102.02515","source_kind":"arxiv","source_version":7}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:0f5f328258738c36203f1026234c15f5070c302014e2d152b4659e5f4f91b52d","sha256:11fa298f11eae4dca88ea9533d137b46351a42f22bedf562a0a3a4fd722ea0cb"],"state_sha256":"7adc1aed1fbcd24106b00997a997c46e2e03db048c7bd37ee181110ac7d3b7a9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1McFP1bt7x5xl3z3h6A4PjCBeQAs2HEvHDh+vkpxKEqrRuSxlvahHwo0a40D+L0Mq2y8wq8dwxSt9k2G5fg8AQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-03T22:52:48.387556Z","bundle_sha256":"9dae091c1490e2fef66fd19260054625f5399d2aaac577772d533ebc7113bffd"}}