{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:OAJ4JM5YNAKDDCEF2FZBEC7API","short_pith_number":"pith:OAJ4JM5Y","schema_version":"1.0","canonical_sha256":"7013c4b3b86814318885d172120be07a067e0a4a92473cd0021d8b171f17bc15","source":{"kind":"arxiv","id":"2110.00296","version":2},"attestation_state":"computed","paper":{"title":"Powerpropagation: A sparsity inducing weight reparameterisation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Jonathan Schwarz, Peter E. Latham, Razvan Pascanu, Siddhant M. Jayakumar, Yee Whye Teh","submitted_at":"2021-10-01T10:03:57Z","abstract_excerpt":"The training of sparse neural networks is becoming an increasingly important tool for reducing the computational footprint of models at training and evaluation, as well enabling the effective scaling up of models. Whereas much work over the years has been dedicated to specialised pruning techniques, little attention has been paid to the inherent effect of gradient based training on model sparsity. In this work, we introduce Powerpropagation, a new weight-parameterisation for neural networks that leads to inherently sparse models. Exploiting the behaviour of gradient descent, our method gives r"},"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":"2110.00296","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2021-10-01T10:03:57Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"34fdde7de7395125d05b2184ce413ec70a066c9265c5da92e09873cae0e4c1bf","abstract_canon_sha256":"10b84a80d49f1a6ad00ba538c63250bc43ea46b86c7b7bc5514af2b9aa5f6107"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:20:24.025700Z","signature_b64":"NDaIlv43T15vvl3L13FQ3TeYIo+sBxbsYbMBeYl7GAabkwnxqWdVH8TEmeXFs0C1+qjB2O+zdgWmUCG7QgOFBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7013c4b3b86814318885d172120be07a067e0a4a92473cd0021d8b171f17bc15","last_reissued_at":"2026-07-05T03:20:24.025273Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:20:24.025273Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Powerpropagation: A sparsity inducing weight reparameterisation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"stat.ML","authors_text":"Jonathan Schwarz, Peter E. Latham, Razvan Pascanu, Siddhant M. Jayakumar, Yee Whye Teh","submitted_at":"2021-10-01T10:03:57Z","abstract_excerpt":"The training of sparse neural networks is becoming an increasingly important tool for reducing the computational footprint of models at training and evaluation, as well enabling the effective scaling up of models. Whereas much work over the years has been dedicated to specialised pruning techniques, little attention has been paid to the inherent effect of gradient based training on model sparsity. In this work, we introduce Powerpropagation, a new weight-parameterisation for neural networks that leads to inherently sparse models. Exploiting the behaviour of gradient descent, our method gives r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2110.00296","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/2110.00296/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":"2110.00296","created_at":"2026-07-05T03:20:24.025337+00:00"},{"alias_kind":"arxiv_version","alias_value":"2110.00296v2","created_at":"2026-07-05T03:20:24.025337+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2110.00296","created_at":"2026-07-05T03:20:24.025337+00:00"},{"alias_kind":"pith_short_12","alias_value":"OAJ4JM5YNAKD","created_at":"2026-07-05T03:20:24.025337+00:00"},{"alias_kind":"pith_short_16","alias_value":"OAJ4JM5YNAKDDCEF","created_at":"2026-07-05T03:20:24.025337+00:00"},{"alias_kind":"pith_short_8","alias_value":"OAJ4JM5Y","created_at":"2026-07-05T03:20:24.025337+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.12224","citing_title":"Optimizers Qualitatively Alter Solutions And We Should Leverage This","ref_index":68,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API","json":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API.json","graph_json":"https://pith.science/api/pith-number/OAJ4JM5YNAKDDCEF2FZBEC7API/graph.json","events_json":"https://pith.science/api/pith-number/OAJ4JM5YNAKDDCEF2FZBEC7API/events.json","paper":"https://pith.science/paper/OAJ4JM5Y"},"agent_actions":{"view_html":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API","download_json":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API.json","view_paper":"https://pith.science/paper/OAJ4JM5Y","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2110.00296&json=true","fetch_graph":"https://pith.science/api/pith-number/OAJ4JM5YNAKDDCEF2FZBEC7API/graph.json","fetch_events":"https://pith.science/api/pith-number/OAJ4JM5YNAKDDCEF2FZBEC7API/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API/action/storage_attestation","attest_author":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API/action/author_attestation","sign_citation":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API/action/citation_signature","submit_replication":"https://pith.science/pith/OAJ4JM5YNAKDDCEF2FZBEC7API/action/replication_record"}},"created_at":"2026-07-05T03:20:24.025337+00:00","updated_at":"2026-07-05T03:20:24.025337+00:00"}