{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:QQTZPFJXYYNQYOMPAFCWM7W6RU","short_pith_number":"pith:QQTZPFJX","schema_version":"1.0","canonical_sha256":"8427979537c61b0c398f0145667ede8d3ceefa0e8eb86d4a45422b116e3b1c71","source":{"kind":"arxiv","id":"2506.16884","version":2},"attestation_state":"computed","paper":{"title":"The Importance of Being Lazy: Scaling Limits of Continual Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alessandro Breccia, Giulia Lanzillotta, Jacopo Graldi, Lorenzo Noci, Thomas Hofmann","submitted_at":"2025-06-20T10:12:38Z","abstract_excerpt":"Despite recent efforts, neural networks still struggle to learn in non-stationary environments, and our understanding of catastrophic forgetting (CF) is far from complete. In this work, we perform a systematic study on the impact of model scale and the degree of feature learning in continual learning. We reconcile existing contradictory observations on scale in the literature, by differentiating between lazy and rich training regimes through a variable parameterization of the architecture. We show that increasing model width is only beneficial when it reduces the amount of feature learning, yi"},"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":"2506.16884","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-06-20T10:12:38Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"40a13be51fc05002b9779214c53f1dbda35ade138f1ebfe5701d25ec60f3b51c","abstract_canon_sha256":"068940600a616527f158d87de199e5c60dce9b23f0ce999a41747076b2263fce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:53:04.544712Z","signature_b64":"DZMhIl9vzmwyi54yxsJrrLoFAbKAHC4TZ5MYIdQGGzX36aTsBkRVP/GlAAIAQwV4YY4VqEh4s3XcbppevS1zBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8427979537c61b0c398f0145667ede8d3ceefa0e8eb86d4a45422b116e3b1c71","last_reissued_at":"2026-07-05T11:53:04.544206Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:53:04.544206Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Importance of Being Lazy: Scaling Limits of Continual Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Alessandro Breccia, Giulia Lanzillotta, Jacopo Graldi, Lorenzo Noci, Thomas Hofmann","submitted_at":"2025-06-20T10:12:38Z","abstract_excerpt":"Despite recent efforts, neural networks still struggle to learn in non-stationary environments, and our understanding of catastrophic forgetting (CF) is far from complete. In this work, we perform a systematic study on the impact of model scale and the degree of feature learning in continual learning. We reconcile existing contradictory observations on scale in the literature, by differentiating between lazy and rich training regimes through a variable parameterization of the architecture. We show that increasing model width is only beneficial when it reduces the amount of feature learning, yi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.16884","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/2506.16884/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":"2506.16884","created_at":"2026-07-05T11:53:04.544271+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.16884v2","created_at":"2026-07-05T11:53:04.544271+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.16884","created_at":"2026-07-05T11:53:04.544271+00:00"},{"alias_kind":"pith_short_12","alias_value":"QQTZPFJXYYNQ","created_at":"2026-07-05T11:53:04.544271+00:00"},{"alias_kind":"pith_short_16","alias_value":"QQTZPFJXYYNQYOMP","created_at":"2026-07-05T11:53:04.544271+00:00"},{"alias_kind":"pith_short_8","alias_value":"QQTZPFJX","created_at":"2026-07-05T11:53:04.544271+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.21691","citing_title":"There Will Be a Scientific Theory of Deep Learning","ref_index":97,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21280","citing_title":"ImageHD: Energy-Efficient On-Device Continual Learning of Visual Representations via Hyperdimensional Computing","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU","json":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU.json","graph_json":"https://pith.science/api/pith-number/QQTZPFJXYYNQYOMPAFCWM7W6RU/graph.json","events_json":"https://pith.science/api/pith-number/QQTZPFJXYYNQYOMPAFCWM7W6RU/events.json","paper":"https://pith.science/paper/QQTZPFJX"},"agent_actions":{"view_html":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU","download_json":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU.json","view_paper":"https://pith.science/paper/QQTZPFJX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.16884&json=true","fetch_graph":"https://pith.science/api/pith-number/QQTZPFJXYYNQYOMPAFCWM7W6RU/graph.json","fetch_events":"https://pith.science/api/pith-number/QQTZPFJXYYNQYOMPAFCWM7W6RU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU/action/storage_attestation","attest_author":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU/action/author_attestation","sign_citation":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU/action/citation_signature","submit_replication":"https://pith.science/pith/QQTZPFJXYYNQYOMPAFCWM7W6RU/action/replication_record"}},"created_at":"2026-07-05T11:53:04.544271+00:00","updated_at":"2026-07-05T11:53:04.544271+00:00"}