{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TXJNI7UXOOJCPLZ7TZQJ7WZEMY","short_pith_number":"pith:TXJNI7UX","schema_version":"1.0","canonical_sha256":"9dd2d47e97739227af3f9e609fdb24663922276ba5df90b2e850148a692a2313","source":{"kind":"arxiv","id":"2606.11244","version":1},"attestation_state":"computed","paper":{"title":"SPEAR: A System for Post-Quantization Error-Adaptive Recovery Enabling Efficient Low-Bit LLM Serving","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.AR","authors_text":"Guosheng Hu, Hongyuan Liu, Junming Shao, Qinli Yang, Yawei Li, Zhiqiang Que","submitted_at":"2026-06-04T22:38:53Z","abstract_excerpt":"Efficient large language model (LLM) serving is increasingly constrained by deployment cost. Quantization is a key technique for reducing serving cost, yet even state-of-the-art 4-bit quantizers exhibit a noticeable quality gap from FP16, particularly for smaller models where low-bit serving is most beneficial. We identify a fundamental cause of this gap: quantization error is highly input-dependent and varies substantially across tokens, while existing post-quantization compensation methods are static and apply identical corrections to all inputs. As a result, easy tokens are over-corrected w"},"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":"2606.11244","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2026-06-04T22:38:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"53a82636115a434647eff5e6377e6898003a4d0c22f005f959371d1621ef54b8","abstract_canon_sha256":"01f30ef06c0fb13155b810793aa0689737f24a979c8a83d384485365a513025f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-11T00:08:14.621827Z","signature_b64":"7AVLrUMQWPKEO0FowGTfi3eYyOeNrXY8AHYhKDdXF5L0OiqU3eKhFc9akTpTY5N6hyahlsrfrKUbuFFJfwRoBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9dd2d47e97739227af3f9e609fdb24663922276ba5df90b2e850148a692a2313","last_reissued_at":"2026-06-11T00:08:14.620981Z","signature_status":"signed_v1","first_computed_at":"2026-06-11T00:08:14.620981Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SPEAR: A System for Post-Quantization Error-Adaptive Recovery Enabling Efficient Low-Bit LLM Serving","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.AR","authors_text":"Guosheng Hu, Hongyuan Liu, Junming Shao, Qinli Yang, Yawei Li, Zhiqiang Que","submitted_at":"2026-06-04T22:38:53Z","abstract_excerpt":"Efficient large language model (LLM) serving is increasingly constrained by deployment cost. Quantization is a key technique for reducing serving cost, yet even state-of-the-art 4-bit quantizers exhibit a noticeable quality gap from FP16, particularly for smaller models where low-bit serving is most beneficial. We identify a fundamental cause of this gap: quantization error is highly input-dependent and varies substantially across tokens, while existing post-quantization compensation methods are static and apply identical corrections to all inputs. As a result, easy tokens are over-corrected w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.11244","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/2606.11244/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":"2606.11244","created_at":"2026-06-11T00:08:14.621119+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.11244v1","created_at":"2026-06-11T00:08:14.621119+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.11244","created_at":"2026-06-11T00:08:14.621119+00:00"},{"alias_kind":"pith_short_12","alias_value":"TXJNI7UXOOJC","created_at":"2026-06-11T00:08:14.621119+00:00"},{"alias_kind":"pith_short_16","alias_value":"TXJNI7UXOOJCPLZ7","created_at":"2026-06-11T00:08:14.621119+00:00"},{"alias_kind":"pith_short_8","alias_value":"TXJNI7UX","created_at":"2026-06-11T00:08:14.621119+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/TXJNI7UXOOJCPLZ7TZQJ7WZEMY","json":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY.json","graph_json":"https://pith.science/api/pith-number/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/graph.json","events_json":"https://pith.science/api/pith-number/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/events.json","paper":"https://pith.science/paper/TXJNI7UX"},"agent_actions":{"view_html":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY","download_json":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY.json","view_paper":"https://pith.science/paper/TXJNI7UX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.11244&json=true","fetch_graph":"https://pith.science/api/pith-number/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/graph.json","fetch_events":"https://pith.science/api/pith-number/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/action/storage_attestation","attest_author":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/action/author_attestation","sign_citation":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/action/citation_signature","submit_replication":"https://pith.science/pith/TXJNI7UXOOJCPLZ7TZQJ7WZEMY/action/replication_record"}},"created_at":"2026-06-11T00:08:14.621119+00:00","updated_at":"2026-06-11T00:08:14.621119+00:00"}