{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:67ZCAXBHMQYNXTP32SXVR7VUOT","short_pith_number":"pith:67ZCAXBH","schema_version":"1.0","canonical_sha256":"f7f2205c276430dbcdfbd4af58feb474e44ea25c78ac8f39fe81c500a2cf82d8","source":{"kind":"arxiv","id":"2508.14302","version":2},"attestation_state":"computed","paper":{"title":"GLASS: Global-Local Aggregation for Inference-time Sparsification of LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Amirhossein Rajabpour, Amirmohsen Sattarifard, Chao Gao, Fengyu Sun, Hanlin Xu, Kunlin Zhang, Negar Hassanpour, Sepehr Lavasani","submitted_at":"2025-08-19T22:50:20Z","abstract_excerpt":"Inference-time sparsification is a promising path to deploy large language models (LLMs) on resource-constrained devices, yet existing training-free methods typically estimate feedforward network (FFN) neuron importance from the input prompt alone. We show this prompt-only signal is often unreliable, especially for short prompts and long-form decoding, leading to inaccurate masks and degraded generation fidelity. We propose GLASS, a plug-and-play, training-free framework that stabilizes dynamic FFN pruning by aggregating two complementary views of neuron criticality: local prompt-specific acti"},"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":"2508.14302","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-08-19T22:50:20Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"9b7ea7d3dce266ba76e40fb71447bda7c767c246e5a5e62d68e54be85e620081","abstract_canon_sha256":"9567ffa7bcd4da6abc29b6898d2e08c06a1f962758487cfc3c72ce964d2a8962"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:10:12.407037Z","signature_b64":"59yRAZvrJlas/C7WG6sVdrugPCiwrLro3N120OB2SpDTW3fErqA6hCfxqFcr0TU0qI376mCGh7sAwNQy5vjUBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f7f2205c276430dbcdfbd4af58feb474e44ea25c78ac8f39fe81c500a2cf82d8","last_reissued_at":"2026-05-18T03:10:12.406498Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:10:12.406498Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GLASS: Global-Local Aggregation for Inference-time Sparsification of LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Amirhossein Rajabpour, Amirmohsen Sattarifard, Chao Gao, Fengyu Sun, Hanlin Xu, Kunlin Zhang, Negar Hassanpour, Sepehr Lavasani","submitted_at":"2025-08-19T22:50:20Z","abstract_excerpt":"Inference-time sparsification is a promising path to deploy large language models (LLMs) on resource-constrained devices, yet existing training-free methods typically estimate feedforward network (FFN) neuron importance from the input prompt alone. We show this prompt-only signal is often unreliable, especially for short prompts and long-form decoding, leading to inaccurate masks and degraded generation fidelity. We propose GLASS, a plug-and-play, training-free framework that stabilizes dynamic FFN pruning by aggregating two complementary views of neuron criticality: local prompt-specific acti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.14302","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":""},"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":"2508.14302","created_at":"2026-05-18T03:10:12.406579+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.14302v2","created_at":"2026-05-18T03:10:12.406579+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.14302","created_at":"2026-05-18T03:10:12.406579+00:00"},{"alias_kind":"pith_short_12","alias_value":"67ZCAXBHMQYN","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"67ZCAXBHMQYNXTP3","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"67ZCAXBH","created_at":"2026-05-18T12:33:37.589309+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/67ZCAXBHMQYNXTP32SXVR7VUOT","json":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT.json","graph_json":"https://pith.science/api/pith-number/67ZCAXBHMQYNXTP32SXVR7VUOT/graph.json","events_json":"https://pith.science/api/pith-number/67ZCAXBHMQYNXTP32SXVR7VUOT/events.json","paper":"https://pith.science/paper/67ZCAXBH"},"agent_actions":{"view_html":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT","download_json":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT.json","view_paper":"https://pith.science/paper/67ZCAXBH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.14302&json=true","fetch_graph":"https://pith.science/api/pith-number/67ZCAXBHMQYNXTP32SXVR7VUOT/graph.json","fetch_events":"https://pith.science/api/pith-number/67ZCAXBHMQYNXTP32SXVR7VUOT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT/action/storage_attestation","attest_author":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT/action/author_attestation","sign_citation":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT/action/citation_signature","submit_replication":"https://pith.science/pith/67ZCAXBHMQYNXTP32SXVR7VUOT/action/replication_record"}},"created_at":"2026-05-18T03:10:12.406579+00:00","updated_at":"2026-05-18T03:10:12.406579+00:00"}