{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:75KJ6YBQOAGX3LF256F23YRBUE","short_pith_number":"pith:75KJ6YBQ","schema_version":"1.0","canonical_sha256":"ff549f6030700d7dacbaef8bade221a10a252ff59d8ec880398f6880dd7d3afe","source":{"kind":"arxiv","id":"2408.01803","version":2},"attestation_state":"computed","paper":{"title":"STBLLM: Breaking the 1-Bit Barrier with Structured Binary LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Dayou Du, Lujun Li, Peijie Dong, Qiang Wang, Ruibo Fan, Wei Xue, Xiaowen Chu, Yike Guo, Yuedong Zhong, Yuhan Chen, Zhenheng Tang","submitted_at":"2024-08-03T15:07:44Z","abstract_excerpt":"In this paper, we present the first structural binarization method for LLM compression to less than 1-bit precision. Although LLMs have achieved remarkable performance, their memory-bound nature during the inference stage hinders the adoption of resource-constrained devices. Reducing weights to 1-bit precision through binarization substantially enhances computational efficiency. We observe that some weights in binarized LLMs can be randomly flipped without significant performance degradation, suggesting the potential for further compression. To exploit this, our STBLLM employs an N:M sparsity "},"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":"2408.01803","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-08-03T15:07:44Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"38211a68d354feb33641686544d9fbef174b337da8e5621a234a00d76c5a82ad","abstract_canon_sha256":"091e984d5b861e1d093b07f9df53e7f811f246a2462c878eeff9862b627d3d6a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:17:14.877222Z","signature_b64":"jJk13LElT+ZCZ5X1prel5A3jSEK0w05aBUVpydrs2aevKlizUuofg+sdAIABwu2IjwGRikdWJSNls8spbnXYAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff549f6030700d7dacbaef8bade221a10a252ff59d8ec880398f6880dd7d3afe","last_reissued_at":"2026-07-05T09:17:14.876813Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:17:14.876813Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"STBLLM: Breaking the 1-Bit Barrier with Structured Binary LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Dayou Du, Lujun Li, Peijie Dong, Qiang Wang, Ruibo Fan, Wei Xue, Xiaowen Chu, Yike Guo, Yuedong Zhong, Yuhan Chen, Zhenheng Tang","submitted_at":"2024-08-03T15:07:44Z","abstract_excerpt":"In this paper, we present the first structural binarization method for LLM compression to less than 1-bit precision. Although LLMs have achieved remarkable performance, their memory-bound nature during the inference stage hinders the adoption of resource-constrained devices. Reducing weights to 1-bit precision through binarization substantially enhances computational efficiency. We observe that some weights in binarized LLMs can be randomly flipped without significant performance degradation, suggesting the potential for further compression. To exploit this, our STBLLM employs an N:M sparsity "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.01803","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2408.01803/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":"2408.01803","created_at":"2026-07-05T09:17:14.876870+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.01803v2","created_at":"2026-07-05T09:17:14.876870+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.01803","created_at":"2026-07-05T09:17:14.876870+00:00"},{"alias_kind":"pith_short_12","alias_value":"75KJ6YBQOAGX","created_at":"2026-07-05T09:17:14.876870+00:00"},{"alias_kind":"pith_short_16","alias_value":"75KJ6YBQOAGX3LF2","created_at":"2026-07-05T09:17:14.876870+00:00"},{"alias_kind":"pith_short_8","alias_value":"75KJ6YBQ","created_at":"2026-07-05T09:17:14.876870+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01876","citing_title":"SAB-LVLM: Significance-Aware Binarization for Large Vision-Language Models","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2512.21651","citing_title":"Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18556","citing_title":"GSQ: Highly-Accurate Low-Precision Scalar Quantization for LLMs via Gumbel-Softmax Sampling","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2512.21651","citing_title":"Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2602.20309","citing_title":"QuantVLA: Scale-Calibrated Post-Training Quantization for Vision-Language-Action Models","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18556","citing_title":"GSQ: Highly-Accurate Low-Precision Scalar Quantization for LLMs via Gumbel-Softmax Sampling","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE","json":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE.json","graph_json":"https://pith.science/api/pith-number/75KJ6YBQOAGX3LF256F23YRBUE/graph.json","events_json":"https://pith.science/api/pith-number/75KJ6YBQOAGX3LF256F23YRBUE/events.json","paper":"https://pith.science/paper/75KJ6YBQ"},"agent_actions":{"view_html":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE","download_json":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE.json","view_paper":"https://pith.science/paper/75KJ6YBQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.01803&json=true","fetch_graph":"https://pith.science/api/pith-number/75KJ6YBQOAGX3LF256F23YRBUE/graph.json","fetch_events":"https://pith.science/api/pith-number/75KJ6YBQOAGX3LF256F23YRBUE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE/action/storage_attestation","attest_author":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE/action/author_attestation","sign_citation":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE/action/citation_signature","submit_replication":"https://pith.science/pith/75KJ6YBQOAGX3LF256F23YRBUE/action/replication_record"}},"created_at":"2026-07-05T09:17:14.876870+00:00","updated_at":"2026-07-05T09:17:14.876870+00:00"}