{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7OYALTYKOIH7ZNEUXHZ4LSX2BF","short_pith_number":"pith:7OYALTYK","schema_version":"1.0","canonical_sha256":"fbb005cf0a720ffcb494b9f3c5cafa0941a669168b083ff7a98688642ae8f249","source":{"kind":"arxiv","id":"2507.07996","version":1},"attestation_state":"computed","paper":{"title":"Skip a Layer or Loop it? Test-Time Depth Adaptation of Pretrained LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Tianyi Zhou, Yang Li, Ziyue Li","submitted_at":"2025-07-10T17:59:53Z","abstract_excerpt":"Can a pretrained neural network adapt its architecture to different inputs without any finetuning? Do we need all layers for simple tasks, and are they adequate for challenging tasks? We found that the layers of a pretrained large language model (LLM) can be manipulated as separate modules to build a better and even shallower model customized for each test sample. In particular, each layer from the pretrained model can be skipped/pruned or repeated multiple times as recurrent neural networks (RNN), and stacked with others in arbitrary orders, yielding a chain-of-layers (CoLa) per sample. This "},"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":"2507.07996","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-07-10T17:59:53Z","cross_cats_sorted":[],"title_canon_sha256":"5682cb5298bacf7b2acfa93df6f652f24842dd73ad9458cf755595690d3190b3","abstract_canon_sha256":"24753c27da64dc98458d9e9a6926f687d87497aa438c94476971537e05251bb4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:35:08.372568Z","signature_b64":"a3U5SDf9T+7fwqk+uXpelhpOtM/XYe29lLz5nI80AK6dsrdrqx0Zw9e7mR5yEG8hZIFkXgwFGUaSHPmCPLpgDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fbb005cf0a720ffcb494b9f3c5cafa0941a669168b083ff7a98688642ae8f249","last_reissued_at":"2026-07-05T11:35:08.372080Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:35:08.372080Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Skip a Layer or Loop it? Test-Time Depth Adaptation of Pretrained LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Tianyi Zhou, Yang Li, Ziyue Li","submitted_at":"2025-07-10T17:59:53Z","abstract_excerpt":"Can a pretrained neural network adapt its architecture to different inputs without any finetuning? Do we need all layers for simple tasks, and are they adequate for challenging tasks? We found that the layers of a pretrained large language model (LLM) can be manipulated as separate modules to build a better and even shallower model customized for each test sample. In particular, each layer from the pretrained model can be skipped/pruned or repeated multiple times as recurrent neural networks (RNN), and stacked with others in arbitrary orders, yielding a chain-of-layers (CoLa) per sample. This "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.07996","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/2507.07996/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":"2507.07996","created_at":"2026-07-05T11:35:08.372139+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.07996v1","created_at":"2026-07-05T11:35:08.372139+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.07996","created_at":"2026-07-05T11:35:08.372139+00:00"},{"alias_kind":"pith_short_12","alias_value":"7OYALTYKOIH7","created_at":"2026-07-05T11:35:08.372139+00:00"},{"alias_kind":"pith_short_16","alias_value":"7OYALTYKOIH7ZNEU","created_at":"2026-07-05T11:35:08.372139+00:00"},{"alias_kind":"pith_short_8","alias_value":"7OYALTYK","created_at":"2026-07-05T11:35:08.372139+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06574","citing_title":"Skip a Layer or Loop It? Learning Program-of-Layers in LLMs","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2510.12773","citing_title":"Dr.LLM: Dynamic Layer Routing in LLMs","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF","json":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF.json","graph_json":"https://pith.science/api/pith-number/7OYALTYKOIH7ZNEUXHZ4LSX2BF/graph.json","events_json":"https://pith.science/api/pith-number/7OYALTYKOIH7ZNEUXHZ4LSX2BF/events.json","paper":"https://pith.science/paper/7OYALTYK"},"agent_actions":{"view_html":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF","download_json":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF.json","view_paper":"https://pith.science/paper/7OYALTYK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.07996&json=true","fetch_graph":"https://pith.science/api/pith-number/7OYALTYKOIH7ZNEUXHZ4LSX2BF/graph.json","fetch_events":"https://pith.science/api/pith-number/7OYALTYKOIH7ZNEUXHZ4LSX2BF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF/action/storage_attestation","attest_author":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF/action/author_attestation","sign_citation":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF/action/citation_signature","submit_replication":"https://pith.science/pith/7OYALTYKOIH7ZNEUXHZ4LSX2BF/action/replication_record"}},"created_at":"2026-07-05T11:35:08.372139+00:00","updated_at":"2026-07-05T11:35:08.372139+00:00"}