{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:S5S2HYF4BIKGDRHS5OZYHJO6OD","short_pith_number":"pith:S5S2HYF4","schema_version":"1.0","canonical_sha256":"9765a3e0bc0a1461c4f2ebb383a5de70ecda812356ec90d235957e048dd79a3d","source":{"kind":"arxiv","id":"2203.08368","version":5},"attestation_state":"computed","paper":{"title":"Mixed-Precision Neural Network Quantization via Learned Layer-wise Importance","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Chen Tang, Kai Ouyang, Wen Ji, Wenwu Zhu, Yaowei Wang, Yifei Zhu, Zhi Wang","submitted_at":"2022-03-16T03:23:50Z","abstract_excerpt":"The exponentially large discrete search space in mixed-precision quantization (MPQ) makes it hard to determine the optimal bit-width for each layer. Previous works usually resort to iterative search methods on the training set, which consume hundreds or even thousands of GPU-hours. In this study, we reveal that some unique learnable parameters in quantization, namely the scale factors in the quantizer, can serve as importance indicators of a layer, reflecting the contribution of that layer to the final accuracy at certain bit-widths. These importance indicators naturally perceive the numerical"},"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":"2203.08368","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-03-16T03:23:50Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"dfed15f4dda0545db750089626b0af26e96c57935afe76a3e74564b7abfec90e","abstract_canon_sha256":"bbb3907bfffec761ded8f8c2a07cddada8fbd7ab0db1e928b392659e2fb1c746"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:47:57.991577Z","signature_b64":"0R5yTIlWb/al5rvtuks31yQW6YRZ1bguAgUNGRxaRwRtMeR25+sqLPE+/c0pHjOeEZyvFmD7qpwf12TqS9p/CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9765a3e0bc0a1461c4f2ebb383a5de70ecda812356ec90d235957e048dd79a3d","last_reissued_at":"2026-07-05T05:47:57.991041Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:47:57.991041Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mixed-Precision Neural Network Quantization via Learned Layer-wise Importance","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Chen Tang, Kai Ouyang, Wen Ji, Wenwu Zhu, Yaowei Wang, Yifei Zhu, Zhi Wang","submitted_at":"2022-03-16T03:23:50Z","abstract_excerpt":"The exponentially large discrete search space in mixed-precision quantization (MPQ) makes it hard to determine the optimal bit-width for each layer. Previous works usually resort to iterative search methods on the training set, which consume hundreds or even thousands of GPU-hours. In this study, we reveal that some unique learnable parameters in quantization, namely the scale factors in the quantizer, can serve as importance indicators of a layer, reflecting the contribution of that layer to the final accuracy at certain bit-widths. These importance indicators naturally perceive the numerical"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.08368","kind":"arxiv","version":5},"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/2203.08368/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":"2203.08368","created_at":"2026-07-05T05:47:57.991099+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.08368v5","created_at":"2026-07-05T05:47:57.991099+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.08368","created_at":"2026-07-05T05:47:57.991099+00:00"},{"alias_kind":"pith_short_12","alias_value":"S5S2HYF4BIKG","created_at":"2026-07-05T05:47:57.991099+00:00"},{"alias_kind":"pith_short_16","alias_value":"S5S2HYF4BIKGDRHS","created_at":"2026-07-05T05:47:57.991099+00:00"},{"alias_kind":"pith_short_8","alias_value":"S5S2HYF4","created_at":"2026-07-05T05:47:57.991099+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.19481","citing_title":"Win Fast or Lose Slow: Balancing Speed and Accuracy in Latency-Sensitive Decisions of LLMs","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD","json":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD.json","graph_json":"https://pith.science/api/pith-number/S5S2HYF4BIKGDRHS5OZYHJO6OD/graph.json","events_json":"https://pith.science/api/pith-number/S5S2HYF4BIKGDRHS5OZYHJO6OD/events.json","paper":"https://pith.science/paper/S5S2HYF4"},"agent_actions":{"view_html":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD","download_json":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD.json","view_paper":"https://pith.science/paper/S5S2HYF4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.08368&json=true","fetch_graph":"https://pith.science/api/pith-number/S5S2HYF4BIKGDRHS5OZYHJO6OD/graph.json","fetch_events":"https://pith.science/api/pith-number/S5S2HYF4BIKGDRHS5OZYHJO6OD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD/action/storage_attestation","attest_author":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD/action/author_attestation","sign_citation":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD/action/citation_signature","submit_replication":"https://pith.science/pith/S5S2HYF4BIKGDRHS5OZYHJO6OD/action/replication_record"}},"created_at":"2026-07-05T05:47:57.991099+00:00","updated_at":"2026-07-05T05:47:57.991099+00:00"}