{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RMZWACZWZOYRC6QKTQE3SZGLYW","short_pith_number":"pith:RMZWACZW","schema_version":"1.0","canonical_sha256":"8b33600b36cbb1117a0a9c09b964cbc5aefb822bbd279e3ba9f3f4591e2e5859","source":{"kind":"arxiv","id":"2507.18073","version":1},"attestation_state":"computed","paper":{"title":"Squeeze10-LLM: Squeezing LLMs' Weights by 10 Times via a Staged Mixed-Precision Quantization Method","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Baochang Zhang, Haodong Zhu, Juan Zhang, Linlin Yang, Mingbao Lin, Qingcheng Zhu, Runqi Wang, Sheng Xu, Yangyang Ren, Yanjing Li, Yuguang Yang, Zichao Feng","submitted_at":"2025-07-24T03:55:19Z","abstract_excerpt":"Deploying large language models (LLMs) is challenging due to their massive parameters and high computational costs. Ultra low-bit quantization can significantly reduce storage and accelerate inference, but extreme compression (i.e., mean bit-width <= 2) often leads to severe performance degradation. To address this, we propose Squeeze10-LLM, effectively \"squeezing\" 16-bit LLMs' weights by 10 times. Specifically, Squeeze10-LLM is a staged mixed-precision post-training quantization (PTQ) framework and achieves an average of 1.6 bits per weight by quantizing 80% of the weights to 1 bit and 20% to"},"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":"2507.18073","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-07-24T03:55:19Z","cross_cats_sorted":[],"title_canon_sha256":"caf5afb5e52f2880caca7cc1a13226af72e2c776613dbe551a338a13fa49db9c","abstract_canon_sha256":"26f82dec325f52542be95aa81571157a1dda9405de7cee640a4eb22ec17287d6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:42:40.886477Z","signature_b64":"Jp3iEapYIcuHWjFazlxA2xPbbhub+uODJgLCJzo6P6VkmHHP3mFSSk6TxpohAdlz0CdgyzHytacd8/y/yKdpAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8b33600b36cbb1117a0a9c09b964cbc5aefb822bbd279e3ba9f3f4591e2e5859","last_reissued_at":"2026-07-05T11:42:40.885982Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:42:40.885982Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Squeeze10-LLM: Squeezing LLMs' Weights by 10 Times via a Staged Mixed-Precision Quantization Method","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Baochang Zhang, Haodong Zhu, Juan Zhang, Linlin Yang, Mingbao Lin, Qingcheng Zhu, Runqi Wang, Sheng Xu, Yangyang Ren, Yanjing Li, Yuguang Yang, Zichao Feng","submitted_at":"2025-07-24T03:55:19Z","abstract_excerpt":"Deploying large language models (LLMs) is challenging due to their massive parameters and high computational costs. Ultra low-bit quantization can significantly reduce storage and accelerate inference, but extreme compression (i.e., mean bit-width <= 2) often leads to severe performance degradation. To address this, we propose Squeeze10-LLM, effectively \"squeezing\" 16-bit LLMs' weights by 10 times. Specifically, Squeeze10-LLM is a staged mixed-precision post-training quantization (PTQ) framework and achieves an average of 1.6 bits per weight by quantizing 80% of the weights to 1 bit and 20% to"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.18073","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/2507.18073/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":"2507.18073","created_at":"2026-07-05T11:42:40.886042+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.18073v1","created_at":"2026-07-05T11:42:40.886042+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.18073","created_at":"2026-07-05T11:42:40.886042+00:00"},{"alias_kind":"pith_short_12","alias_value":"RMZWACZWZOYR","created_at":"2026-07-05T11:42:40.886042+00:00"},{"alias_kind":"pith_short_16","alias_value":"RMZWACZWZOYRC6QK","created_at":"2026-07-05T11:42:40.886042+00:00"},{"alias_kind":"pith_short_8","alias_value":"RMZWACZW","created_at":"2026-07-05T11:42:40.886042+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08643","citing_title":"BiSCo-LLM: Lookup-Free Binary Spherical Coding for Extreme Low-Bit Large Language Model Compression","ref_index":25,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW","json":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW.json","graph_json":"https://pith.science/api/pith-number/RMZWACZWZOYRC6QKTQE3SZGLYW/graph.json","events_json":"https://pith.science/api/pith-number/RMZWACZWZOYRC6QKTQE3SZGLYW/events.json","paper":"https://pith.science/paper/RMZWACZW"},"agent_actions":{"view_html":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW","download_json":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW.json","view_paper":"https://pith.science/paper/RMZWACZW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.18073&json=true","fetch_graph":"https://pith.science/api/pith-number/RMZWACZWZOYRC6QKTQE3SZGLYW/graph.json","fetch_events":"https://pith.science/api/pith-number/RMZWACZWZOYRC6QKTQE3SZGLYW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW/action/storage_attestation","attest_author":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW/action/author_attestation","sign_citation":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW/action/citation_signature","submit_replication":"https://pith.science/pith/RMZWACZWZOYRC6QKTQE3SZGLYW/action/replication_record"}},"created_at":"2026-07-05T11:42:40.886042+00:00","updated_at":"2026-07-05T11:42:40.886042+00:00"}