{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:J63V3UCTXDZU6KGD6IWSW3AQP6","short_pith_number":"pith:J63V3UCT","schema_version":"1.0","canonical_sha256":"4fb75dd053b8f34f28c3f22d2b6c107f9ed0d693b89b3a8c876954ca727668d7","source":{"kind":"arxiv","id":"2512.04746","version":2},"attestation_state":"computed","paper":{"title":"SignRoundV2: Toward Closing the Performance Gap in Extremely Low-Bit Post-Training Quantization for LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Haihao Shen, Heng Guo, Weiwei Zhang, Wenhua Cheng, Zaner Ma","submitted_at":"2025-12-04T12:35:10Z","abstract_excerpt":"Extremely low-bit quantization is critical for efficiently deploying Large Language Models (LLMs), yet it often leads to severe performance degradation at 2 bits and even at 4 bits (e.g., MXFP4). We present SignRoundV2, a post-training quantization framework designed to maintain high performance even under aggressive compression. SignRoundV2 introduces (1) a simple yet efficient adaptive mixed-precision strategy that leverages gradient information and quantization-induced reconstruction errors to guide layer-wise bit allocation, and (2) a set of lightweight stabilization techniques, including "},"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":"2512.04746","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-12-04T12:35:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7f36e941c2ad3c42c92538c1dea34d70c02557f786485add93e0d574136c6e72","abstract_canon_sha256":"658e4d49a05f09217499bb41180f0d3bccec1481e2f97fc513b42d6daefe35d4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:04:18.472025Z","signature_b64":"jJ+7QE9xmx5xsUoOd/cWCIKTmzZ+JCsZlrerO3WWgla4PCbsHtJiiIG/Gsu7S0SNGy3WDYhZQjNR313pdBacBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4fb75dd053b8f34f28c3f22d2b6c107f9ed0d693b89b3a8c876954ca727668d7","last_reissued_at":"2026-05-20T00:04:18.471173Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:04:18.471173Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SignRoundV2: Toward Closing the Performance Gap in Extremely Low-Bit Post-Training Quantization for LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Haihao Shen, Heng Guo, Weiwei Zhang, Wenhua Cheng, Zaner Ma","submitted_at":"2025-12-04T12:35:10Z","abstract_excerpt":"Extremely low-bit quantization is critical for efficiently deploying Large Language Models (LLMs), yet it often leads to severe performance degradation at 2 bits and even at 4 bits (e.g., MXFP4). We present SignRoundV2, a post-training quantization framework designed to maintain high performance even under aggressive compression. SignRoundV2 introduces (1) a simple yet efficient adaptive mixed-precision strategy that leverages gradient information and quantization-induced reconstruction errors to guide layer-wise bit allocation, and (2) a set of lightweight stabilization techniques, including "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.04746","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/2512.04746/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":"2512.04746","created_at":"2026-05-20T00:04:18.471308+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.04746v2","created_at":"2026-05-20T00:04:18.471308+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.04746","created_at":"2026-05-20T00:04:18.471308+00:00"},{"alias_kind":"pith_short_12","alias_value":"J63V3UCTXDZU","created_at":"2026-05-20T00:04:18.471308+00:00"},{"alias_kind":"pith_short_16","alias_value":"J63V3UCTXDZU6KGD","created_at":"2026-05-20T00:04:18.471308+00:00"},{"alias_kind":"pith_short_8","alias_value":"J63V3UCT","created_at":"2026-05-20T00:04:18.471308+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.18475","citing_title":"GAMMA: Global Bit Allocation for Mixed-Precision Models under Arbitrary Budgets","ref_index":5,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6","json":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6.json","graph_json":"https://pith.science/api/pith-number/J63V3UCTXDZU6KGD6IWSW3AQP6/graph.json","events_json":"https://pith.science/api/pith-number/J63V3UCTXDZU6KGD6IWSW3AQP6/events.json","paper":"https://pith.science/paper/J63V3UCT"},"agent_actions":{"view_html":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6","download_json":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6.json","view_paper":"https://pith.science/paper/J63V3UCT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.04746&json=true","fetch_graph":"https://pith.science/api/pith-number/J63V3UCTXDZU6KGD6IWSW3AQP6/graph.json","fetch_events":"https://pith.science/api/pith-number/J63V3UCTXDZU6KGD6IWSW3AQP6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6/action/storage_attestation","attest_author":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6/action/author_attestation","sign_citation":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6/action/citation_signature","submit_replication":"https://pith.science/pith/J63V3UCTXDZU6KGD6IWSW3AQP6/action/replication_record"}},"created_at":"2026-05-20T00:04:18.471308+00:00","updated_at":"2026-05-20T00:04:18.471308+00:00"}