{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:TOAWZZWEJH36NKXJCWKF2IZUE5","short_pith_number":"pith:TOAWZZWE","schema_version":"1.0","canonical_sha256":"9b816ce6c449f7e6aae915945d2334276e9c5aeb0b8ed67b0b16c3b8e39bd834","source":{"kind":"arxiv","id":"2412.07174","version":1},"attestation_state":"computed","paper":{"title":"Post-Training Statistical Calibration for Higher Activation Sparsity","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Nilesh Jain, Vui Seng Chua, Yujie Pan","submitted_at":"2024-12-10T04:15:39Z","abstract_excerpt":"We present Statistical Calibrated Activation Pruning (SCAP), a post-training activation pruning framework that (1) generalizes sparsification by input activations of Fully-Connected layers for generic and flexible application across Transformers, and (2) features a simple Mode-Centering technique to pre-calibrate activation distributions for maximizing post-training sparsity. Our results demonstrate robust Pareto efficiency compared to prior methods, translating to a 1.5x additional LLM decoding speedup against CATS at iso model quality. SCAP effectiveness is empirically verified across a wide"},"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":"2412.07174","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-12-10T04:15:39Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"02ae81dd832c61188c7cfa1f05a532aa8ce90f714551c3913cdff80533d6f5b3","abstract_canon_sha256":"fb36ddc03df0dc222f20e648bea2844f68c27f369da07ba12ee1fc291468c36d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:47:05.145717Z","signature_b64":"EGDh8SBm5qD88eH5S5EoD7tR9ux3yLJkrB3aKQ0aRr90SoyU2fO9gZO0wtc4XdOPDBRviPAMSanvGmJbaMB/Ag==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9b816ce6c449f7e6aae915945d2334276e9c5aeb0b8ed67b0b16c3b8e39bd834","last_reissued_at":"2026-07-05T09:47:05.145264Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:47:05.145264Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Post-Training Statistical Calibration for Higher Activation Sparsity","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Nilesh Jain, Vui Seng Chua, Yujie Pan","submitted_at":"2024-12-10T04:15:39Z","abstract_excerpt":"We present Statistical Calibrated Activation Pruning (SCAP), a post-training activation pruning framework that (1) generalizes sparsification by input activations of Fully-Connected layers for generic and flexible application across Transformers, and (2) features a simple Mode-Centering technique to pre-calibrate activation distributions for maximizing post-training sparsity. Our results demonstrate robust Pareto efficiency compared to prior methods, translating to a 1.5x additional LLM decoding speedup against CATS at iso model quality. SCAP effectiveness is empirically verified across a wide"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.07174","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/2412.07174/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":"2412.07174","created_at":"2026-07-05T09:47:05.145322+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.07174v1","created_at":"2026-07-05T09:47:05.145322+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.07174","created_at":"2026-07-05T09:47:05.145322+00:00"},{"alias_kind":"pith_short_12","alias_value":"TOAWZZWEJH36","created_at":"2026-07-05T09:47:05.145322+00:00"},{"alias_kind":"pith_short_16","alias_value":"TOAWZZWEJH36NKXJ","created_at":"2026-07-05T09:47:05.145322+00:00"},{"alias_kind":"pith_short_8","alias_value":"TOAWZZWE","created_at":"2026-07-05T09:47:05.145322+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.17659","citing_title":"Bug or Feature$^2$: Weight Drift, Activation Sparsity and Spikes","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17659","citing_title":"Bug or Feature$^2$: Weight Drift, Activation Sparsity and Spikes","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2509.22166","citing_title":"Motivating Next-Gen Accelerators with Flexible (N:M) Activation Sparsity via Benchmarking Lightweight Post-Training Sparsification Approaches","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5","json":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5.json","graph_json":"https://pith.science/api/pith-number/TOAWZZWEJH36NKXJCWKF2IZUE5/graph.json","events_json":"https://pith.science/api/pith-number/TOAWZZWEJH36NKXJCWKF2IZUE5/events.json","paper":"https://pith.science/paper/TOAWZZWE"},"agent_actions":{"view_html":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5","download_json":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5.json","view_paper":"https://pith.science/paper/TOAWZZWE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.07174&json=true","fetch_graph":"https://pith.science/api/pith-number/TOAWZZWEJH36NKXJCWKF2IZUE5/graph.json","fetch_events":"https://pith.science/api/pith-number/TOAWZZWEJH36NKXJCWKF2IZUE5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5/action/storage_attestation","attest_author":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5/action/author_attestation","sign_citation":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5/action/citation_signature","submit_replication":"https://pith.science/pith/TOAWZZWEJH36NKXJCWKF2IZUE5/action/replication_record"}},"created_at":"2026-07-05T09:47:05.145322+00:00","updated_at":"2026-07-05T09:47:05.145322+00:00"}