{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6W26OJRHLUCFE2AMTERPKULRVX","short_pith_number":"pith:6W26OJRH","schema_version":"1.0","canonical_sha256":"f5b5e726275d0452680c9922f55171adc2483d2369ecdf8792d507765db223ef","source":{"kind":"arxiv","id":"2504.14152","version":1},"attestation_state":"computed","paper":{"title":"FGMP: Fine-Grained Mixed-Precision Weight and Activation Quantization for Hardware-Accelerated LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AR","authors_text":"Ben Keller, Brucek Khailany, Charbel Sakr, Coleman Hooper, Kurt Keutzer, Rangharajan Venkatesan, Sophia Shao","submitted_at":"2025-04-19T02:51:45Z","abstract_excerpt":"Quantization is a powerful tool to improve large language model (LLM) inference efficiency by utilizing more energy-efficient low-precision datapaths and reducing memory footprint. However, accurately quantizing LLM weights and activations to low precision is challenging without degrading model accuracy. We propose fine-grained mixed precision (FGMP) quantization, a post-training mixed-precision quantization hardware-software co-design methodology that maintains accuracy while quantizing the majority of weights and activations to reduced precision. Our work makes the following contributions: 1"},"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":"2504.14152","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2025-04-19T02:51:45Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"74e0fa9571517e3ecb1ffe76edb106829709e1a5575ccb065c14e5cc7e629950","abstract_canon_sha256":"e8c593ef7a3f4a564283ec98ce6caea17a298d2921e7ace48a7fdd949e910b36"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:51:11.838521Z","signature_b64":"IRISfmPchh8kWnl4vjhsgg5zjqOkGpJPxS+B8s+kdXFjOxzqJOPJv6vm9fuYbGYg2WveJW0LUqZVMvcn4+GxCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f5b5e726275d0452680c9922f55171adc2483d2369ecdf8792d507765db223ef","last_reissued_at":"2026-07-05T10:51:11.837789Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:51:11.837789Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FGMP: Fine-Grained Mixed-Precision Weight and Activation Quantization for Hardware-Accelerated LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AR","authors_text":"Ben Keller, Brucek Khailany, Charbel Sakr, Coleman Hooper, Kurt Keutzer, Rangharajan Venkatesan, Sophia Shao","submitted_at":"2025-04-19T02:51:45Z","abstract_excerpt":"Quantization is a powerful tool to improve large language model (LLM) inference efficiency by utilizing more energy-efficient low-precision datapaths and reducing memory footprint. However, accurately quantizing LLM weights and activations to low precision is challenging without degrading model accuracy. We propose fine-grained mixed precision (FGMP) quantization, a post-training mixed-precision quantization hardware-software co-design methodology that maintains accuracy while quantizing the majority of weights and activations to reduced precision. Our work makes the following contributions: 1"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.14152","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/2504.14152/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":"2504.14152","created_at":"2026-07-05T10:51:11.837860+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.14152v1","created_at":"2026-07-05T10:51:11.837860+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.14152","created_at":"2026-07-05T10:51:11.837860+00:00"},{"alias_kind":"pith_short_12","alias_value":"6W26OJRHLUCF","created_at":"2026-07-05T10:51:11.837860+00:00"},{"alias_kind":"pith_short_16","alias_value":"6W26OJRHLUCFE2AM","created_at":"2026-07-05T10:51:11.837860+00:00"},{"alias_kind":"pith_short_8","alias_value":"6W26OJRH","created_at":"2026-07-05T10:51:11.837860+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26587","citing_title":"SharQ: Bridging Activation Sparsity and FP4 Quantization for LLM Inference","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX","json":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX.json","graph_json":"https://pith.science/api/pith-number/6W26OJRHLUCFE2AMTERPKULRVX/graph.json","events_json":"https://pith.science/api/pith-number/6W26OJRHLUCFE2AMTERPKULRVX/events.json","paper":"https://pith.science/paper/6W26OJRH"},"agent_actions":{"view_html":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX","download_json":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX.json","view_paper":"https://pith.science/paper/6W26OJRH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.14152&json=true","fetch_graph":"https://pith.science/api/pith-number/6W26OJRHLUCFE2AMTERPKULRVX/graph.json","fetch_events":"https://pith.science/api/pith-number/6W26OJRHLUCFE2AMTERPKULRVX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX/action/storage_attestation","attest_author":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX/action/author_attestation","sign_citation":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX/action/citation_signature","submit_replication":"https://pith.science/pith/6W26OJRHLUCFE2AMTERPKULRVX/action/replication_record"}},"created_at":"2026-07-05T10:51:11.837860+00:00","updated_at":"2026-07-05T10:51:11.837860+00:00"}