{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:CHGU24OWAT36YUGVWSFJCAYPU6","short_pith_number":"pith:CHGU24OW","schema_version":"1.0","canonical_sha256":"11cd4d71d604f7ec50d5b48a91030fa7bcc84a3e5be919279c0750c9d1c34e85","source":{"kind":"arxiv","id":"2506.04179","version":1},"attestation_state":"computed","paper":{"title":"SkipGPT: Dynamic Layer Pruning Reinvented with Token Awareness and Module Decoupling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anhao Zhao, Fanghua Ye, Hui Su, Junlong Tong, Xiaoyu Shen, Yingqi Fan, Zhiwei Fei","submitted_at":"2025-06-04T17:26:31Z","abstract_excerpt":"Large language models (LLMs) achieve remarkable performance across tasks but incur substantial computational costs due to their deep, multi-layered architectures. Layer pruning has emerged as a strategy to alleviate these inefficiencies, but conventional static pruning methods overlook two critical dynamics inherent to LLM inference: (1) horizontal dynamics, where token-level heterogeneity demands context-aware pruning decisions, and (2) vertical dynamics, where the distinct functional roles of MLP and self-attention layers necessitate component-specific pruning policies. We introduce SkipGPT,"},"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":"2506.04179","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-04T17:26:31Z","cross_cats_sorted":[],"title_canon_sha256":"958cc40c8c16a9ab18dcf7fe3432b627a4adddc8403ed6e67757400e7e7e7211","abstract_canon_sha256":"af310bd2f3b19f580ea3773c68a9251713b251bc62331b5ac8f16a6c06b95d42"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:15:58.408148Z","signature_b64":"IxQGIkXwUQNHXwzVFu8+PD94NAp6ixSM1msT7SyNGFvMNcDInNH1syeN8aH2fxTDhKAAkrF5hmo9IaxaGYBABg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"11cd4d71d604f7ec50d5b48a91030fa7bcc84a3e5be919279c0750c9d1c34e85","last_reissued_at":"2026-07-05T11:15:58.407735Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:15:58.407735Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SkipGPT: Dynamic Layer Pruning Reinvented with Token Awareness and Module Decoupling","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anhao Zhao, Fanghua Ye, Hui Su, Junlong Tong, Xiaoyu Shen, Yingqi Fan, Zhiwei Fei","submitted_at":"2025-06-04T17:26:31Z","abstract_excerpt":"Large language models (LLMs) achieve remarkable performance across tasks but incur substantial computational costs due to their deep, multi-layered architectures. Layer pruning has emerged as a strategy to alleviate these inefficiencies, but conventional static pruning methods overlook two critical dynamics inherent to LLM inference: (1) horizontal dynamics, where token-level heterogeneity demands context-aware pruning decisions, and (2) vertical dynamics, where the distinct functional roles of MLP and self-attention layers necessitate component-specific pruning policies. We introduce SkipGPT,"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.04179","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/2506.04179/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":"2506.04179","created_at":"2026-07-05T11:15:58.407791+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.04179v1","created_at":"2026-07-05T11:15:58.407791+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.04179","created_at":"2026-07-05T11:15:58.407791+00:00"},{"alias_kind":"pith_short_12","alias_value":"CHGU24OWAT36","created_at":"2026-07-05T11:15:58.407791+00:00"},{"alias_kind":"pith_short_16","alias_value":"CHGU24OWAT36YUGV","created_at":"2026-07-05T11:15:58.407791+00:00"},{"alias_kind":"pith_short_8","alias_value":"CHGU24OW","created_at":"2026-07-05T11:15:58.407791+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27743","citing_title":"End-to-End Dynamic Sparsity for Resource-Adaptive LLM Inference","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00523","citing_title":"ProactiveLLM: Learning Active Interaction for Streaming Large Language Models","ref_index":115,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26538","citing_title":"CascadeFormer: Depth-Tapered Transformers Motivated by Gradient Fan-in Asymmetry","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01167","citing_title":"Do All Individual Layers Help? An Empirical Study of Task-Interfering Layers in Vision-Language Models","ref_index":57,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6","json":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6.json","graph_json":"https://pith.science/api/pith-number/CHGU24OWAT36YUGVWSFJCAYPU6/graph.json","events_json":"https://pith.science/api/pith-number/CHGU24OWAT36YUGVWSFJCAYPU6/events.json","paper":"https://pith.science/paper/CHGU24OW"},"agent_actions":{"view_html":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6","download_json":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6.json","view_paper":"https://pith.science/paper/CHGU24OW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.04179&json=true","fetch_graph":"https://pith.science/api/pith-number/CHGU24OWAT36YUGVWSFJCAYPU6/graph.json","fetch_events":"https://pith.science/api/pith-number/CHGU24OWAT36YUGVWSFJCAYPU6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6/action/storage_attestation","attest_author":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6/action/author_attestation","sign_citation":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6/action/citation_signature","submit_replication":"https://pith.science/pith/CHGU24OWAT36YUGVWSFJCAYPU6/action/replication_record"}},"created_at":"2026-07-05T11:15:58.407791+00:00","updated_at":"2026-07-05T11:15:58.407791+00:00"}