{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2020:WWGCH5TNU2EMINUOR7QXS4ESTS","short_pith_number":"pith:WWGCH5TN","canonical_record":{"source":{"id":"2004.15012","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-30T17:56:30Z","cross_cats_sorted":[],"title_canon_sha256":"cbc0db88abdac61a7cc6ab88177c0148b1ce3932769df68d7f3c834af5e56416","abstract_canon_sha256":"e188c899f634239398d4c7860089df42009bcf395da59f3ada25403c223042fa"},"schema_version":"1.0"},"canonical_sha256":"b58c23f66da688c4368e8fe17970929ca08da8a2e113d10271809c0e245cd948","source":{"kind":"arxiv","id":"2004.15012","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2004.15012","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"arxiv_version","alias_value":"2004.15012v2","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.15012","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"pith_short_12","alias_value":"WWGCH5TNU2EM","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"pith_short_16","alias_value":"WWGCH5TNU2EMINUO","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"pith_short_8","alias_value":"WWGCH5TN","created_at":"2026-07-05T01:41:31Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2020:WWGCH5TNU2EMINUOR7QXS4ESTS","target":"record","payload":{"canonical_record":{"source":{"id":"2004.15012","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-30T17:56:30Z","cross_cats_sorted":[],"title_canon_sha256":"cbc0db88abdac61a7cc6ab88177c0148b1ce3932769df68d7f3c834af5e56416","abstract_canon_sha256":"e188c899f634239398d4c7860089df42009bcf395da59f3ada25403c223042fa"},"schema_version":"1.0"},"canonical_sha256":"b58c23f66da688c4368e8fe17970929ca08da8a2e113d10271809c0e245cd948","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:41:31.049042Z","signature_b64":"TV+aS69pKhj/kD+HtTelfq5IeVK8RDAfdKA1hfzlAvYqt04pqZVzpItyYqqUNLPqL3GSwgewLxXwPWX7SCMaCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b58c23f66da688c4368e8fe17970929ca08da8a2e113d10271809c0e245cd948","last_reissued_at":"2026-07-05T01:41:31.048673Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:41:31.048673Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2004.15012","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T01:41:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ruygm+fGj6J0mmDYTp5j6gb+hiCV28dW3qNfkRuldU48wN0sSNK6G/uVd4HWwcb28cGUD8O5VN/3Zz+KsYnsDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-06T14:52:23.535208Z"},"content_sha256":"9fd7563e1a0c38ed54072e96de4ed6d22213b357510c843506241771d2f0381a","schema_version":"1.0","event_id":"sha256:9fd7563e1a0c38ed54072e96de4ed6d22213b357510c843506241771d2f0381a"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2020:WWGCH5TNU2EMINUOR7QXS4ESTS","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Does Data Augmentation Improve Generalization in NLP?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Charles Lovering, Ellie Pavlick, Rohan Jha","submitted_at":"2020-04-30T17:56:30Z","abstract_excerpt":"Neural models often exploit superficial features to achieve good performance, rather than deriving more general features. Overcoming this tendency is a central challenge in areas such as representation learning and ML fairness. Recent work has proposed using data augmentation, i.e., generating training examples where the superficial features fail, as a means of encouraging models to prefer the stronger features. We design a series of toy learning problems to test the hypothesis that data augmentation leads models to unlearn weaker heuristics, but not to learn stronger features in their place. "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.15012","kind":"arxiv","version":2},"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/2004.15012/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":"2026-07-05T01:41:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QzrcP97Cqci8PGLXJDnos6YrCIi1cbEmQ4zTJ7fy1CMMPf4r/2fxvrMXEvNLn1D2ogKNJAtQ6zyViw1u7uQZCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-06T14:52:23.535596Z"},"content_sha256":"282e6ea6a109e313f5c237eb76e19c36fa96719aa194181f67f4865e99b83d13","schema_version":"1.0","event_id":"sha256:282e6ea6a109e313f5c237eb76e19c36fa96719aa194181f67f4865e99b83d13"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/WWGCH5TNU2EMINUOR7QXS4ESTS/bundle.json","state_url":"https://pith.science/pith/WWGCH5TNU2EMINUOR7QXS4ESTS/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/WWGCH5TNU2EMINUOR7QXS4ESTS/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":"2026-07-06T14:52:23Z","links":{"resolver":"https://pith.science/pith/WWGCH5TNU2EMINUOR7QXS4ESTS","bundle":"https://pith.science/pith/WWGCH5TNU2EMINUOR7QXS4ESTS/bundle.json","state":"https://pith.science/pith/WWGCH5TNU2EMINUOR7QXS4ESTS/state.json","well_known_bundle":"https://pith.science/.well-known/pith/WWGCH5TNU2EMINUOR7QXS4ESTS/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2020:WWGCH5TNU2EMINUOR7QXS4ESTS","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":"e188c899f634239398d4c7860089df42009bcf395da59f3ada25403c223042fa","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-30T17:56:30Z","title_canon_sha256":"cbc0db88abdac61a7cc6ab88177c0148b1ce3932769df68d7f3c834af5e56416"},"schema_version":"1.0","source":{"id":"2004.15012","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2004.15012","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"arxiv_version","alias_value":"2004.15012v2","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.15012","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"pith_short_12","alias_value":"WWGCH5TNU2EM","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"pith_short_16","alias_value":"WWGCH5TNU2EMINUO","created_at":"2026-07-05T01:41:31Z"},{"alias_kind":"pith_short_8","alias_value":"WWGCH5TN","created_at":"2026-07-05T01:41:31Z"}],"graph_snapshots":[{"event_id":"sha256:282e6ea6a109e313f5c237eb76e19c36fa96719aa194181f67f4865e99b83d13","target":"graph","created_at":"2026-07-05T01:41:31Z","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/2004.15012/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Neural models often exploit superficial features to achieve good performance, rather than deriving more general features. Overcoming this tendency is a central challenge in areas such as representation learning and ML fairness. Recent work has proposed using data augmentation, i.e., generating training examples where the superficial features fail, as a means of encouraging models to prefer the stronger features. We design a series of toy learning problems to test the hypothesis that data augmentation leads models to unlearn weaker heuristics, but not to learn stronger features in their place. ","authors_text":"Charles Lovering, Ellie Pavlick, Rohan Jha","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-30T17:56:30Z","title":"Does Data Augmentation Improve Generalization in NLP?"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.15012","kind":"arxiv","version":2},"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":"sha256:9fd7563e1a0c38ed54072e96de4ed6d22213b357510c843506241771d2f0381a","target":"record","created_at":"2026-07-05T01:41:31Z","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":"e188c899f634239398d4c7860089df42009bcf395da59f3ada25403c223042fa","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-30T17:56:30Z","title_canon_sha256":"cbc0db88abdac61a7cc6ab88177c0148b1ce3932769df68d7f3c834af5e56416"},"schema_version":"1.0","source":{"id":"2004.15012","kind":"arxiv","version":2}},"canonical_sha256":"b58c23f66da688c4368e8fe17970929ca08da8a2e113d10271809c0e245cd948","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b58c23f66da688c4368e8fe17970929ca08da8a2e113d10271809c0e245cd948","first_computed_at":"2026-07-05T01:41:31.048673Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T01:41:31.048673Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"TV+aS69pKhj/kD+HtTelfq5IeVK8RDAfdKA1hfzlAvYqt04pqZVzpItyYqqUNLPqL3GSwgewLxXwPWX7SCMaCQ==","signature_status":"signed_v1","signed_at":"2026-07-05T01:41:31.049042Z","signed_message":"canonical_sha256_bytes"},"source_id":"2004.15012","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:9fd7563e1a0c38ed54072e96de4ed6d22213b357510c843506241771d2f0381a","sha256:282e6ea6a109e313f5c237eb76e19c36fa96719aa194181f67f4865e99b83d13"],"state_sha256":"1a6b4f6e5f002cc9df388262f41ffe35f92fdec9983f4de4fe4784f9f1652f74"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"I0D4xxXBwmLBCCIOHyu52NxQqG5K7DDsSmJqmtrDuiJ4aNyfN1HPhRuTDNPfBdzvGVuE+QOapoUxR/LCmHOLBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-06T14:52:23.537592Z","bundle_sha256":"a3005ab4206a6188e8306801636dc63cd8baefbd000dcaa052fdc9d0fe01c487"}}