{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2017:5NIRUFXP3X4RDY7XD7ZDFPRG7L","short_pith_number":"pith:5NIRUFXP","schema_version":"1.0","canonical_sha256":"eb511a16efddf911e3f71ff232be26fafc32c062d79cdb93a7933048f8e649e8","source":{"kind":"arxiv","id":"1706.08840","version":6},"attestation_state":"computed","paper":{"title":"Gradient Episodic Memory for Continual Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"David Lopez-Paz, Marc'Aurelio Ranzato","submitted_at":"2017-06-26T14:53:34Z","abstract_excerpt":"One major obstacle towards AI is the poor ability of models to solve new problems quicker, and without forgetting previously acquired knowledge. To better understand this issue, we study the problem of continual learning, where the model observes, once and one by one, examples concerning a sequence of tasks. First, we propose a set of metrics to evaluate models learning over a continuum of data. These metrics characterize models not only by their test accuracy, but also in terms of their ability to transfer knowledge across tasks. Second, we propose a model for continual learning, called Gradi"},"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":"1706.08840","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2017-06-26T14:53:34Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1bfdc09eee423d23337d391934aaf767f0e53cdea32f40d01f5120d3dfcefbff","abstract_canon_sha256":"7d4ca3f9f29e8dc9d6bf733e76f1572b1bdf794a639ab407cd49d39dd149577f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:56:57.442849Z","signature_b64":"ebFnXCvmbU1KPj0wcjcegllaMlt0Tqy6Mx5M3aeZVjqRzsC19m9D/jWF+1yCoo8ehVDb3xC7jKF19Z6h3YQtAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eb511a16efddf911e3f71ff232be26fafc32c062d79cdb93a7933048f8e649e8","last_reissued_at":"2026-07-05T04:56:57.442380Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:56:57.442380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gradient Episodic Memory for Continual Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"David Lopez-Paz, Marc'Aurelio Ranzato","submitted_at":"2017-06-26T14:53:34Z","abstract_excerpt":"One major obstacle towards AI is the poor ability of models to solve new problems quicker, and without forgetting previously acquired knowledge. To better understand this issue, we study the problem of continual learning, where the model observes, once and one by one, examples concerning a sequence of tasks. First, we propose a set of metrics to evaluate models learning over a continuum of data. These metrics characterize models not only by their test accuracy, but also in terms of their ability to transfer knowledge across tasks. Second, we propose a model for continual learning, called Gradi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1706.08840","kind":"arxiv","version":6},"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/1706.08840/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":"1706.08840","created_at":"2026-07-05T04:56:57.442444+00:00"},{"alias_kind":"arxiv_version","alias_value":"1706.08840v6","created_at":"2026-07-05T04:56:57.442444+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1706.08840","created_at":"2026-07-05T04:56:57.442444+00:00"},{"alias_kind":"pith_short_12","alias_value":"5NIRUFXP3X4R","created_at":"2026-07-05T04:56:57.442444+00:00"},{"alias_kind":"pith_short_16","alias_value":"5NIRUFXP3X4RDY7X","created_at":"2026-07-05T04:56:57.442444+00:00"},{"alias_kind":"pith_short_8","alias_value":"5NIRUFXP","created_at":"2026-07-05T04:56:57.442444+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01988","citing_title":"Episodic-to-Semantic Consolidation Without Identity Drift","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01858","citing_title":"Polaris: Scaling Up Instruction-Guided Image Generation Towards Millions of Personalized Style Needs","ref_index":86,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17545","citing_title":"Continuous-time Optimal Stopping through Deep Reinforcement Learning","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2412.02125","citing_title":"Preference Goal Tuning: Post-Training as Latent Control for Frozen Policies","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20296","citing_title":"Spectral Unforgetting: Post-Hoc Recovery of Damaged Capabilities Without Retraining","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21927","citing_title":"Fine-Tuning Regimes Define Distinct Continual Learning Problems","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L","json":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L.json","graph_json":"https://pith.science/api/pith-number/5NIRUFXP3X4RDY7XD7ZDFPRG7L/graph.json","events_json":"https://pith.science/api/pith-number/5NIRUFXP3X4RDY7XD7ZDFPRG7L/events.json","paper":"https://pith.science/paper/5NIRUFXP"},"agent_actions":{"view_html":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L","download_json":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L.json","view_paper":"https://pith.science/paper/5NIRUFXP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1706.08840&json=true","fetch_graph":"https://pith.science/api/pith-number/5NIRUFXP3X4RDY7XD7ZDFPRG7L/graph.json","fetch_events":"https://pith.science/api/pith-number/5NIRUFXP3X4RDY7XD7ZDFPRG7L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L/action/storage_attestation","attest_author":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L/action/author_attestation","sign_citation":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L/action/citation_signature","submit_replication":"https://pith.science/pith/5NIRUFXP3X4RDY7XD7ZDFPRG7L/action/replication_record"}},"created_at":"2026-07-05T04:56:57.442444+00:00","updated_at":"2026-07-05T04:56:57.442444+00:00"}