{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:O5JZXJAQCV5JVJEG5HDCWGQ5KP","short_pith_number":"pith:O5JZXJAQ","schema_version":"1.0","canonical_sha256":"77539ba410157a9aa486e9c62b1a1d53f3585af69eb162b82e74ce58f63fb3d2","source":{"kind":"arxiv","id":"2606.18246","version":1},"attestation_state":"computed","paper":{"title":"Variable-Width Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Oliver Sieberling, Rameswar Panda, Shawn Tan, Yoon Kim, Yury Polyanskiy, Zhaofeng Wu","submitted_at":"2026-06-16T17:59:03Z","abstract_excerpt":"Scaling model size, specifically depth and width, has driven significant progress in transformer-based language models. However, most architectures maintain a constant width across all layers, allocating a fixed parameter and computation budget evenly despite different layers potentially playing distinct computational roles. In this work, we empirically investigate nonuniform capacity allocation across network depth by proposing a $\\times$-shaped > <former architecture. This design maintains wider early and late layers while narrowing the middle layers, utilizing a parameter-free residual resi"},"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":"2606.18246","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-16T17:59:03Z","cross_cats_sorted":[],"title_canon_sha256":"0df607d957d378792306809a70b9054a7a1f313a03ecdc96cfbfa937afe40c67","abstract_canon_sha256":"399a1a2d2a1ded7cebcf2d30ec43f47cf9973999b2943a73a30552a0769943b1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:10:52.438769Z","signature_b64":"Zn2AXmMOCpUh8ibtFQ0QsVSp04DS+SojzYZwXeDnoDp7sEIr7F1P4gMn2yBdni5HLAGlnzUOy1X2jb2q4BS3AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"77539ba410157a9aa486e9c62b1a1d53f3585af69eb162b82e74ce58f63fb3d2","last_reissued_at":"2026-06-19T16:10:52.438420Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:10:52.438420Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Variable-Width Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Oliver Sieberling, Rameswar Panda, Shawn Tan, Yoon Kim, Yury Polyanskiy, Zhaofeng Wu","submitted_at":"2026-06-16T17:59:03Z","abstract_excerpt":"Scaling model size, specifically depth and width, has driven significant progress in transformer-based language models. However, most architectures maintain a constant width across all layers, allocating a fixed parameter and computation budget evenly despite different layers potentially playing distinct computational roles. In this work, we empirically investigate nonuniform capacity allocation across network depth by proposing a $\\times$-shaped > <former architecture. This design maintains wider early and late layers while narrowing the middle layers, utilizing a parameter-free residual resi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.18246","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/2606.18246/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":"2606.18246","created_at":"2026-06-19T16:10:52.438483+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.18246v1","created_at":"2026-06-19T16:10:52.438483+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.18246","created_at":"2026-06-19T16:10:52.438483+00:00"},{"alias_kind":"pith_short_12","alias_value":"O5JZXJAQCV5J","created_at":"2026-06-19T16:10:52.438483+00:00"},{"alias_kind":"pith_short_16","alias_value":"O5JZXJAQCV5JVJEG","created_at":"2026-06-19T16:10:52.438483+00:00"},{"alias_kind":"pith_short_8","alias_value":"O5JZXJAQ","created_at":"2026-06-19T16:10:52.438483+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/O5JZXJAQCV5JVJEG5HDCWGQ5KP","json":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP.json","graph_json":"https://pith.science/api/pith-number/O5JZXJAQCV5JVJEG5HDCWGQ5KP/graph.json","events_json":"https://pith.science/api/pith-number/O5JZXJAQCV5JVJEG5HDCWGQ5KP/events.json","paper":"https://pith.science/paper/O5JZXJAQ"},"agent_actions":{"view_html":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP","download_json":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP.json","view_paper":"https://pith.science/paper/O5JZXJAQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.18246&json=true","fetch_graph":"https://pith.science/api/pith-number/O5JZXJAQCV5JVJEG5HDCWGQ5KP/graph.json","fetch_events":"https://pith.science/api/pith-number/O5JZXJAQCV5JVJEG5HDCWGQ5KP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP/action/storage_attestation","attest_author":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP/action/author_attestation","sign_citation":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP/action/citation_signature","submit_replication":"https://pith.science/pith/O5JZXJAQCV5JVJEG5HDCWGQ5KP/action/replication_record"}},"created_at":"2026-06-19T16:10:52.438483+00:00","updated_at":"2026-06-19T16:10:52.438483+00:00"}