{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:GVGAP36YEWUXA7VTD6S7LMVX2C","short_pith_number":"pith:GVGAP36Y","schema_version":"1.0","canonical_sha256":"354c07efd825a9707eb31fa5f5b2b7d08a1f74ea018ab608ccb4362c9aa06c84","source":{"kind":"arxiv","id":"2412.18790","version":1},"attestation_state":"computed","paper":{"title":"Torque-Aware Momentum","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aristide Baratin, Gintare Karolina Dziugaite, Goncalo Mordido, Pranshu Malviya, Razvan Pascanu, Reza Babanezhad Harikandeh, Sarath Chandar","submitted_at":"2024-12-25T05:58:07Z","abstract_excerpt":"Efficiently exploring complex loss landscapes is key to the performance of deep neural networks. While momentum-based optimizers are widely used in state-of-the-art setups, classical momentum can still struggle with large, misaligned gradients, leading to oscillations. To address this, we propose Torque-Aware Momentum (TAM), which introduces a damping factor based on the angle between the new gradients and previous momentum, stabilizing the update direction during training. Empirical results show that TAM, which can be combined with both SGD and Adam, enhances exploration, handles distribution"},"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":"2412.18790","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-12-25T05:58:07Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"25d2eb557207951b4cb6c11da3c3c067bf2fbacd1c816202beb0bee967eb622d","abstract_canon_sha256":"c29ab73a366346c3123b67cc004aacbeaa114bf995446c64c94c5782f40319d1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:54:10.859172Z","signature_b64":"mZXKohGEgPqFsLEIo5mEjQpECmWbWfWdEOY5+J+36qeBl+D1qkCwxGWniJqdMid54RDIY0WuBMrm7tFkaoRqAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"354c07efd825a9707eb31fa5f5b2b7d08a1f74ea018ab608ccb4362c9aa06c84","last_reissued_at":"2026-07-05T09:54:10.858735Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:54:10.858735Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Torque-Aware Momentum","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aristide Baratin, Gintare Karolina Dziugaite, Goncalo Mordido, Pranshu Malviya, Razvan Pascanu, Reza Babanezhad Harikandeh, Sarath Chandar","submitted_at":"2024-12-25T05:58:07Z","abstract_excerpt":"Efficiently exploring complex loss landscapes is key to the performance of deep neural networks. While momentum-based optimizers are widely used in state-of-the-art setups, classical momentum can still struggle with large, misaligned gradients, leading to oscillations. To address this, we propose Torque-Aware Momentum (TAM), which introduces a damping factor based on the angle between the new gradients and previous momentum, stabilizing the update direction during training. Empirical results show that TAM, which can be combined with both SGD and Adam, enhances exploration, handles distribution"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.18790","kind":"arxiv","version":1},"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/2412.18790/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":"2412.18790","created_at":"2026-07-05T09:54:10.858796+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.18790v1","created_at":"2026-07-05T09:54:10.858796+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.18790","created_at":"2026-07-05T09:54:10.858796+00:00"},{"alias_kind":"pith_short_12","alias_value":"GVGAP36YEWUX","created_at":"2026-07-05T09:54:10.858796+00:00"},{"alias_kind":"pith_short_16","alias_value":"GVGAP36YEWUXA7VT","created_at":"2026-07-05T09:54:10.858796+00:00"},{"alias_kind":"pith_short_8","alias_value":"GVGAP36Y","created_at":"2026-07-05T09:54:10.858796+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C","json":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C.json","graph_json":"https://pith.science/api/pith-number/GVGAP36YEWUXA7VTD6S7LMVX2C/graph.json","events_json":"https://pith.science/api/pith-number/GVGAP36YEWUXA7VTD6S7LMVX2C/events.json","paper":"https://pith.science/paper/GVGAP36Y"},"agent_actions":{"view_html":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C","download_json":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C.json","view_paper":"https://pith.science/paper/GVGAP36Y","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.18790&json=true","fetch_graph":"https://pith.science/api/pith-number/GVGAP36YEWUXA7VTD6S7LMVX2C/graph.json","fetch_events":"https://pith.science/api/pith-number/GVGAP36YEWUXA7VTD6S7LMVX2C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C/action/storage_attestation","attest_author":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C/action/author_attestation","sign_citation":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C/action/citation_signature","submit_replication":"https://pith.science/pith/GVGAP36YEWUXA7VTD6S7LMVX2C/action/replication_record"}},"created_at":"2026-07-05T09:54:10.858796+00:00","updated_at":"2026-07-05T09:54:10.858796+00:00"}