{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:XPMKT3BSQCBFVNP2UV724TLPEW","short_pith_number":"pith:XPMKT3BS","canonical_record":{"source":{"id":"2402.02998","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-05T13:37:00Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"5085be9b1a7421e6acc198d51a5d9740d3db446cae069e8b9135c2c4755732da","abstract_canon_sha256":"f42d2f1640fe8cc5e603c6d6dca578a0982814133de5b2c406547413b5941a17"},"schema_version":"1.0"},"canonical_sha256":"bbd8a9ec3280825ab5faa57fae4d6f25b83bccda645f3e84edc530031c48e4a1","source":{"kind":"arxiv","id":"2402.02998","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.02998","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"arxiv_version","alias_value":"2402.02998v2","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.02998","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"pith_short_12","alias_value":"XPMKT3BSQCBF","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"pith_short_16","alias_value":"XPMKT3BSQCBFVNP2","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"pith_short_8","alias_value":"XPMKT3BS","created_at":"2026-07-05T10:25:04Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:XPMKT3BSQCBFVNP2UV724TLPEW","target":"record","payload":{"canonical_record":{"source":{"id":"2402.02998","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-05T13:37:00Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"5085be9b1a7421e6acc198d51a5d9740d3db446cae069e8b9135c2c4755732da","abstract_canon_sha256":"f42d2f1640fe8cc5e603c6d6dca578a0982814133de5b2c406547413b5941a17"},"schema_version":"1.0"},"canonical_sha256":"bbd8a9ec3280825ab5faa57fae4d6f25b83bccda645f3e84edc530031c48e4a1","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:25:04.510368Z","signature_b64":"xwJmRs9GLDtW86mpIFAWtndoa5Gk3Gr99If/neOJvSmxtxAX3qqLfND92DZK5UNTx7mtkfwJIZm7DvgzivDHCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bbd8a9ec3280825ab5faa57fae4d6f25b83bccda645f3e84edc530031c48e4a1","last_reissued_at":"2026-07-05T10:25:04.509864Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:25:04.509864Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2402.02998","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":"2026-07-05T10:25:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5szrvouzrggmpbalEvI1PLN9Ol6yEBahYBIpSWO66qBwyc6B/oyJo1ESGy7FIjW9RVWprkrtgHmKUnI/m/gXDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-31T18:46:55.764128Z"},"content_sha256":"a160ebc428f8e0d1c6c94008f54b01f87cf1068cf89b1a2e915bcdc63bfec9c5","schema_version":"1.0","event_id":"sha256:a160ebc428f8e0d1c6c94008f54b01f87cf1068cf89b1a2e915bcdc63bfec9c5"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:XPMKT3BSQCBFVNP2UV724TLPEW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Careful with that Scalpel: Improving Gradient Surgery with an EMA","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Eugene Ndiaye, James Thornton, Marco Cuturi, Michal Klein, Pierre Ablin, Yu-Guan Hsieh","submitted_at":"2024-02-05T13:37:00Z","abstract_excerpt":"Beyond minimizing a single training loss, many deep learning estimation pipelines rely on an auxiliary objective to quantify and encourage desirable properties of the model (e.g. performance on another dataset, robustness, agreement with a prior). Although the simplest approach to incorporating an auxiliary loss is to sum it with the training loss as a regularizer, recent works have shown that one can improve performance by blending the gradients beyond a simple sum; this is known as gradient surgery. We cast the problem as a constrained minimization problem where the auxiliary objective is mi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.02998","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/2402.02998/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":"2026-07-05T10:25:04Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mkAhMIAF/eEBaa7mdEQhItKNQMFiFlohvuP2TfHUauKAzaauDbUMAMp+ZhbTG+3jIso680RxKxovJmhwUMOrBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-31T18:46:55.764631Z"},"content_sha256":"9b101793d1a2444fb15456fe818252584857f8946c7b770ec31fd311c0fcc95b","schema_version":"1.0","event_id":"sha256:9b101793d1a2444fb15456fe818252584857f8946c7b770ec31fd311c0fcc95b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XPMKT3BSQCBFVNP2UV724TLPEW/bundle.json","state_url":"https://pith.science/pith/XPMKT3BSQCBFVNP2UV724TLPEW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XPMKT3BSQCBFVNP2UV724TLPEW/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":"2026-07-31T18:46:55Z","links":{"resolver":"https://pith.science/pith/XPMKT3BSQCBFVNP2UV724TLPEW","bundle":"https://pith.science/pith/XPMKT3BSQCBFVNP2UV724TLPEW/bundle.json","state":"https://pith.science/pith/XPMKT3BSQCBFVNP2UV724TLPEW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XPMKT3BSQCBFVNP2UV724TLPEW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:XPMKT3BSQCBFVNP2UV724TLPEW","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":"f42d2f1640fe8cc5e603c6d6dca578a0982814133de5b2c406547413b5941a17","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-05T13:37:00Z","title_canon_sha256":"5085be9b1a7421e6acc198d51a5d9740d3db446cae069e8b9135c2c4755732da"},"schema_version":"1.0","source":{"id":"2402.02998","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2402.02998","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"arxiv_version","alias_value":"2402.02998v2","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.02998","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"pith_short_12","alias_value":"XPMKT3BSQCBF","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"pith_short_16","alias_value":"XPMKT3BSQCBFVNP2","created_at":"2026-07-05T10:25:04Z"},{"alias_kind":"pith_short_8","alias_value":"XPMKT3BS","created_at":"2026-07-05T10:25:04Z"}],"graph_snapshots":[{"event_id":"sha256:9b101793d1a2444fb15456fe818252584857f8946c7b770ec31fd311c0fcc95b","target":"graph","created_at":"2026-07-05T10:25:04Z","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/2402.02998/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Beyond minimizing a single training loss, many deep learning estimation pipelines rely on an auxiliary objective to quantify and encourage desirable properties of the model (e.g. performance on another dataset, robustness, agreement with a prior). Although the simplest approach to incorporating an auxiliary loss is to sum it with the training loss as a regularizer, recent works have shown that one can improve performance by blending the gradients beyond a simple sum; this is known as gradient surgery. We cast the problem as a constrained minimization problem where the auxiliary objective is mi","authors_text":"Eugene Ndiaye, James Thornton, Marco Cuturi, Michal Klein, Pierre Ablin, Yu-Guan Hsieh","cross_cats":["stat.ML"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-05T13:37:00Z","title":"Careful with that Scalpel: Improving Gradient Surgery with an EMA"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.02998","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:a160ebc428f8e0d1c6c94008f54b01f87cf1068cf89b1a2e915bcdc63bfec9c5","target":"record","created_at":"2026-07-05T10:25:04Z","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":"f42d2f1640fe8cc5e603c6d6dca578a0982814133de5b2c406547413b5941a17","cross_cats_sorted":["stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-02-05T13:37:00Z","title_canon_sha256":"5085be9b1a7421e6acc198d51a5d9740d3db446cae069e8b9135c2c4755732da"},"schema_version":"1.0","source":{"id":"2402.02998","kind":"arxiv","version":2}},"canonical_sha256":"bbd8a9ec3280825ab5faa57fae4d6f25b83bccda645f3e84edc530031c48e4a1","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bbd8a9ec3280825ab5faa57fae4d6f25b83bccda645f3e84edc530031c48e4a1","first_computed_at":"2026-07-05T10:25:04.509864Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:25:04.509864Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"xwJmRs9GLDtW86mpIFAWtndoa5Gk3Gr99If/neOJvSmxtxAX3qqLfND92DZK5UNTx7mtkfwJIZm7DvgzivDHCA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:25:04.510368Z","signed_message":"canonical_sha256_bytes"},"source_id":"2402.02998","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a160ebc428f8e0d1c6c94008f54b01f87cf1068cf89b1a2e915bcdc63bfec9c5","sha256:9b101793d1a2444fb15456fe818252584857f8946c7b770ec31fd311c0fcc95b"],"state_sha256":"ce8b2bf6f35e9e20021edebcba8d07da5a79db567a8b39aaaef138e9de5d271f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"EWx1cdLrUQ5Os1TblTrfDB0Bvzx8fPrPoCHb2U1KM8mL5j+UYtsQSQocOfmwhuGMec2ypCGDZ1IoMRDx/J9OAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-31T18:46:55.769749Z","bundle_sha256":"634b9d2309199a71725377a06d0c0bf78efe72ccaee3e81a6d8f49554fe8824c"}}