{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4ZVIP2LDYFUAW255I7CJUFEEVM","short_pith_number":"pith:4ZVIP2LD","schema_version":"1.0","canonical_sha256":"e66a87e963c1680b6bbd47c49a1484ab0ddfdd9d60d0c7eb6ca3d28b5b997c52","source":{"kind":"arxiv","id":"2504.14992","version":2},"attestation_state":"computed","paper":{"title":"Efficient Pretraining Length Scaling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bohong Wu, Jianqiao Lu, Shen Yan, Sijun Zhang, Xun Zhou, Ya Wang, Yutao Zeng","submitted_at":"2025-04-21T09:41:26Z","abstract_excerpt":"Recent advances in large language models have demonstrated the effectiveness of length scaling during post-training, yet its potential in pre-training remains underexplored. We present the Parallel Hidden Decoding Transformer (\\textit{PHD}-Transformer), a novel framework that enables efficient length scaling during pre-training while maintaining inference efficiency. \\textit{PHD}-Transformer achieves this through an innovative KV cache management strategy that distinguishes between original tokens and hidden decoding tokens. By retaining only the KV cache of original tokens for long-range depe"},"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":"2504.14992","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-21T09:41:26Z","cross_cats_sorted":[],"title_canon_sha256":"71fb85d2795fd00f3c8c88c642b6a71632c702e6798c1e04390b13dbb19dd2be","abstract_canon_sha256":"d46ff0005426d580e1a08410843a52f4c3133bb4eb688e918e59b3de30f8cd54"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:53:17.900934Z","signature_b64":"Yr1HKqllJBBjdOVcvQiG+XMfG2e0Bx6a2igxqMY3vZqS/7t8PLP3KhXdyiB1gSAqsvESeQSW2luZwOIPk6q9CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e66a87e963c1680b6bbd47c49a1484ab0ddfdd9d60d0c7eb6ca3d28b5b997c52","last_reissued_at":"2026-07-05T10:53:17.900457Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:53:17.900457Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Efficient Pretraining Length Scaling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bohong Wu, Jianqiao Lu, Shen Yan, Sijun Zhang, Xun Zhou, Ya Wang, Yutao Zeng","submitted_at":"2025-04-21T09:41:26Z","abstract_excerpt":"Recent advances in large language models have demonstrated the effectiveness of length scaling during post-training, yet its potential in pre-training remains underexplored. We present the Parallel Hidden Decoding Transformer (\\textit{PHD}-Transformer), a novel framework that enables efficient length scaling during pre-training while maintaining inference efficiency. \\textit{PHD}-Transformer achieves this through an innovative KV cache management strategy that distinguishes between original tokens and hidden decoding tokens. By retaining only the KV cache of original tokens for long-range depe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.14992","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/2504.14992/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":"2504.14992","created_at":"2026-07-05T10:53:17.900514+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.14992v2","created_at":"2026-07-05T10:53:17.900514+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.14992","created_at":"2026-07-05T10:53:17.900514+00:00"},{"alias_kind":"pith_short_12","alias_value":"4ZVIP2LDYFUA","created_at":"2026-07-05T10:53:17.900514+00:00"},{"alias_kind":"pith_short_16","alias_value":"4ZVIP2LDYFUAW255","created_at":"2026-07-05T10:53:17.900514+00:00"},{"alias_kind":"pith_short_8","alias_value":"4ZVIP2LD","created_at":"2026-07-05T10:53:17.900514+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.25741","citing_title":"Scaling Latent Reasoning via Looped Language Models","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09630","citing_title":"Scratchpad Patching: Decoupling Compute from Patch Size in Byte-Level Language Models","ref_index":92,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM","json":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM.json","graph_json":"https://pith.science/api/pith-number/4ZVIP2LDYFUAW255I7CJUFEEVM/graph.json","events_json":"https://pith.science/api/pith-number/4ZVIP2LDYFUAW255I7CJUFEEVM/events.json","paper":"https://pith.science/paper/4ZVIP2LD"},"agent_actions":{"view_html":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM","download_json":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM.json","view_paper":"https://pith.science/paper/4ZVIP2LD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.14992&json=true","fetch_graph":"https://pith.science/api/pith-number/4ZVIP2LDYFUAW255I7CJUFEEVM/graph.json","fetch_events":"https://pith.science/api/pith-number/4ZVIP2LDYFUAW255I7CJUFEEVM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM/action/storage_attestation","attest_author":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM/action/author_attestation","sign_citation":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM/action/citation_signature","submit_replication":"https://pith.science/pith/4ZVIP2LDYFUAW255I7CJUFEEVM/action/replication_record"}},"created_at":"2026-07-05T10:53:17.900514+00:00","updated_at":"2026-07-05T10:53:17.900514+00:00"}