{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MEOEI6BYTCR2F62URQU7NI7RQY","short_pith_number":"pith:MEOEI6BY","schema_version":"1.0","canonical_sha256":"611c44783898a3a2fb548c29f6a3f1862437f090fd9f87c22369747de628dfa5","source":{"kind":"arxiv","id":"2410.05871","version":2},"attestation_state":"computed","paper":{"title":"A second-order-like optimizer with adaptive gradient scaling for deep learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Andrei Purica, Edouard Pauwels (TSE-R, IRIT-ADRIA), J\\'er\\^ome Bolte (TSE-R), Ryan Boustany (TSE-R)","submitted_at":"2024-10-08T09:58:38Z","abstract_excerpt":"In this empirical article, we introduce INNAprop, an optimization algorithm that combines the INNA method with the RMSprop adaptive gradient scaling. It leverages second-order information and rescaling while keeping the memory requirements of standard DL methods as AdamW or SGD with momentum. After giving geometrical insights, we evaluate INNAprop on CIFAR-10, Food101, and ImageNet with ResNets, VGG, DenseNet, and ViT, and on GPT-2 (OpenWebText) train from scratch and with LoRA fine-tuning (E2E). INNAprop consistently matches or outperforms AdamW both in training speed and accuracy, with minim"},"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":"2410.05871","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-10-08T09:58:38Z","cross_cats_sorted":["cs.AI","math.OC"],"title_canon_sha256":"4f988a541c814299c8974aad64539c568ca6a9ae166fdb0cec1f2caad82421de","abstract_canon_sha256":"4c5681fe7dc5a0fd254302d45f3ccb5daedcb1292f8a0d2ef630fc2e76e3af0d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:48:16.108771Z","signature_b64":"FyqbBgUMy/lPgxIdv+D+tXvj/qZmhgUm3HvqZDcU6B/GWT8iD7NrWNgY4Qfa9I6m8VcdZiatlWQy4u9W357/BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"611c44783898a3a2fb548c29f6a3f1862437f090fd9f87c22369747de628dfa5","last_reissued_at":"2026-07-05T09:48:16.108259Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:48:16.108259Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A second-order-like optimizer with adaptive gradient scaling for deep learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","math.OC"],"primary_cat":"cs.LG","authors_text":"Andrei Purica, Edouard Pauwels (TSE-R, IRIT-ADRIA), J\\'er\\^ome Bolte (TSE-R), Ryan Boustany (TSE-R)","submitted_at":"2024-10-08T09:58:38Z","abstract_excerpt":"In this empirical article, we introduce INNAprop, an optimization algorithm that combines the INNA method with the RMSprop adaptive gradient scaling. It leverages second-order information and rescaling while keeping the memory requirements of standard DL methods as AdamW or SGD with momentum. After giving geometrical insights, we evaluate INNAprop on CIFAR-10, Food101, and ImageNet with ResNets, VGG, DenseNet, and ViT, and on GPT-2 (OpenWebText) train from scratch and with LoRA fine-tuning (E2E). INNAprop consistently matches or outperforms AdamW both in training speed and accuracy, with minim"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.05871","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/2410.05871/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":"2410.05871","created_at":"2026-07-05T09:48:16.108337+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.05871v2","created_at":"2026-07-05T09:48:16.108337+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.05871","created_at":"2026-07-05T09:48:16.108337+00:00"},{"alias_kind":"pith_short_12","alias_value":"MEOEI6BYTCR2","created_at":"2026-07-05T09:48:16.108337+00:00"},{"alias_kind":"pith_short_16","alias_value":"MEOEI6BYTCR2F62U","created_at":"2026-07-05T09:48:16.108337+00:00"},{"alias_kind":"pith_short_8","alias_value":"MEOEI6BY","created_at":"2026-07-05T09:48:16.108337+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2504.20096","citing_title":"Towards Practical Second-Order Optimizers in Deep Learning: Insights from Fisher Information Analysis","ref_index":30,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY","json":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY.json","graph_json":"https://pith.science/api/pith-number/MEOEI6BYTCR2F62URQU7NI7RQY/graph.json","events_json":"https://pith.science/api/pith-number/MEOEI6BYTCR2F62URQU7NI7RQY/events.json","paper":"https://pith.science/paper/MEOEI6BY"},"agent_actions":{"view_html":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY","download_json":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY.json","view_paper":"https://pith.science/paper/MEOEI6BY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.05871&json=true","fetch_graph":"https://pith.science/api/pith-number/MEOEI6BYTCR2F62URQU7NI7RQY/graph.json","fetch_events":"https://pith.science/api/pith-number/MEOEI6BYTCR2F62URQU7NI7RQY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY/action/storage_attestation","attest_author":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY/action/author_attestation","sign_citation":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY/action/citation_signature","submit_replication":"https://pith.science/pith/MEOEI6BYTCR2F62URQU7NI7RQY/action/replication_record"}},"created_at":"2026-07-05T09:48:16.108337+00:00","updated_at":"2026-07-05T09:48:16.108337+00:00"}