{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UQTGCJSAUJSPXJCYG7PNEH6SL4","short_pith_number":"pith:UQTGCJSA","schema_version":"1.0","canonical_sha256":"a426612640a264fba45837ded21fd25f1d525e06a1249ecb5d8c2e4916d01969","source":{"kind":"arxiv","id":"2607.16612","version":1},"attestation_state":"computed","paper":{"title":"Backpropagation-Free Trunk Training via the Split Forward Gradients","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Tian Qin, Wei-Min Huang","submitted_at":"2026-07-18T03:23:21Z","abstract_excerpt":"Backpropagation makes training deep networks memory intensive because it must store intermediate activations. Forward-mode methods avoid this cost, but their gradient estimates become increasingly noisy as the number of trained parameters grows. We introduce Split Forward Gradient (Split-FG), which splits a network at an intermediate representation: it computes the output head gradient exactly and estimates only the trunk gradient with a Jacobian--vector product. This reduces estimator variance and requires no backward pass through the trunk, while retaining an Adam-style convergence guarantee"},"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":"2607.16612","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"stat.ML","submitted_at":"2026-07-18T03:23:21Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"915d679927a4a8831c0dc15b292e4260f550d4154883c62943c183983e450586","abstract_canon_sha256":"e3943626dec509642e695889f1c45100d23afc04d33cf06190f3db5280bd530e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-21T00:20:28.298986Z","signature_b64":"gTZboE4MBnlc25xv92K4GnI3lVSkWMXDNm881px2kWXauneK4bp4AR7hZBPjjJB6sUFfCQOTsl/6Msti4WhIBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a426612640a264fba45837ded21fd25f1d525e06a1249ecb5d8c2e4916d01969","last_reissued_at":"2026-07-21T00:20:28.298094Z","signature_status":"signed_v1","first_computed_at":"2026-07-21T00:20:28.298094Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Backpropagation-Free Trunk Training via the Split Forward Gradients","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"stat.ML","authors_text":"Tian Qin, Wei-Min Huang","submitted_at":"2026-07-18T03:23:21Z","abstract_excerpt":"Backpropagation makes training deep networks memory intensive because it must store intermediate activations. Forward-mode methods avoid this cost, but their gradient estimates become increasingly noisy as the number of trained parameters grows. We introduce Split Forward Gradient (Split-FG), which splits a network at an intermediate representation: it computes the output head gradient exactly and estimates only the trunk gradient with a Jacobian--vector product. This reduces estimator variance and requires no backward pass through the trunk, while retaining an Adam-style convergence guarantee"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.16612","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/2607.16612/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":"2607.16612","created_at":"2026-07-21T00:20:28.298564+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.16612v1","created_at":"2026-07-21T00:20:28.298564+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.16612","created_at":"2026-07-21T00:20:28.298564+00:00"},{"alias_kind":"pith_short_12","alias_value":"UQTGCJSAUJSP","created_at":"2026-07-21T00:20:28.298564+00:00"},{"alias_kind":"pith_short_16","alias_value":"UQTGCJSAUJSPXJCY","created_at":"2026-07-21T00:20:28.298564+00:00"},{"alias_kind":"pith_short_8","alias_value":"UQTGCJSA","created_at":"2026-07-21T00:20:28.298564+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/UQTGCJSAUJSPXJCYG7PNEH6SL4","json":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4.json","graph_json":"https://pith.science/api/pith-number/UQTGCJSAUJSPXJCYG7PNEH6SL4/graph.json","events_json":"https://pith.science/api/pith-number/UQTGCJSAUJSPXJCYG7PNEH6SL4/events.json","paper":"https://pith.science/paper/UQTGCJSA"},"agent_actions":{"view_html":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4","download_json":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4.json","view_paper":"https://pith.science/paper/UQTGCJSA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.16612&json=true","fetch_graph":"https://pith.science/api/pith-number/UQTGCJSAUJSPXJCYG7PNEH6SL4/graph.json","fetch_events":"https://pith.science/api/pith-number/UQTGCJSAUJSPXJCYG7PNEH6SL4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4/action/storage_attestation","attest_author":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4/action/author_attestation","sign_citation":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4/action/citation_signature","submit_replication":"https://pith.science/pith/UQTGCJSAUJSPXJCYG7PNEH6SL4/action/replication_record"}},"created_at":"2026-07-21T00:20:28.298564+00:00","updated_at":"2026-07-21T00:20:28.298564+00:00"}