{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:OGZ2ETE6LULHZ3GKEDKT6COOQU","short_pith_number":"pith:OGZ2ETE6","schema_version":"1.0","canonical_sha256":"71b3a24c9e5d167cecca20d53f09ce8524041b0e14b8dbf9ab4defb72d8aeff3","source":{"kind":"arxiv","id":"2511.06516","version":3},"attestation_state":"computed","paper":{"title":"You Had One Job: Per-Task Quantization Using LLMs' Hidden Representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amit Levi, Avi Mendelson, Chaim Baskin, Ravid Shwartz Ziv, Raz Lapid, Rom Himelstein","submitted_at":"2025-11-09T19:58:24Z","abstract_excerpt":"Many LLM applications require only narrow capabilities, yet standard post-training quantization (PTQ) methods allocate precision without considering the target task. This can waste bits on layers that are less relevant to the task signal while over-compressing layers that are critical for downstream behavior. We propose Task-Aware Quantization (TAQ), a training-free, weight-only mixed-precision PTQ framework that uses a small set of unlabeled task calibration prompts to allocate higher precision to task-relevant transformer layers under a fixed bit budget. TAQ estimates layer importance from h"},"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":"2511.06516","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-11-09T19:58:24Z","cross_cats_sorted":[],"title_canon_sha256":"5afd45b003e51c5ea29d67ceb440008ecc5218316ebf8ab0954cd0d244448bc3","abstract_canon_sha256":"9528cdfb6adf15f94c9e108dba67b62a3dcfdfada45310b7ce0109a2d1fc60a5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:02:58.276509Z","signature_b64":"NV9UVX8XJRWPsb2IGNwZRLvpILlDqiHpZdbHhW5NiaYqDyA7/cZjGLHLWm8moDYZbRTcVnH3B/90V/fOR7ESCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"71b3a24c9e5d167cecca20d53f09ce8524041b0e14b8dbf9ab4defb72d8aeff3","last_reissued_at":"2026-05-20T00:02:58.275576Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:02:58.275576Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"You Had One Job: Per-Task Quantization Using LLMs' Hidden Representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Amit Levi, Avi Mendelson, Chaim Baskin, Ravid Shwartz Ziv, Raz Lapid, Rom Himelstein","submitted_at":"2025-11-09T19:58:24Z","abstract_excerpt":"Many LLM applications require only narrow capabilities, yet standard post-training quantization (PTQ) methods allocate precision without considering the target task. This can waste bits on layers that are less relevant to the task signal while over-compressing layers that are critical for downstream behavior. We propose Task-Aware Quantization (TAQ), a training-free, weight-only mixed-precision PTQ framework that uses a small set of unlabeled task calibration prompts to allocate higher precision to task-relevant transformer layers under a fixed bit budget. TAQ estimates layer importance from h"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.06516","kind":"arxiv","version":3},"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/2511.06516/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":"2511.06516","created_at":"2026-05-20T00:02:58.275734+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.06516v3","created_at":"2026-05-20T00:02:58.275734+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.06516","created_at":"2026-05-20T00:02:58.275734+00:00"},{"alias_kind":"pith_short_12","alias_value":"OGZ2ETE6LULH","created_at":"2026-05-20T00:02:58.275734+00:00"},{"alias_kind":"pith_short_16","alias_value":"OGZ2ETE6LULHZ3GK","created_at":"2026-05-20T00:02:58.275734+00:00"},{"alias_kind":"pith_short_8","alias_value":"OGZ2ETE6","created_at":"2026-05-20T00:02:58.275734+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU","json":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU.json","graph_json":"https://pith.science/api/pith-number/OGZ2ETE6LULHZ3GKEDKT6COOQU/graph.json","events_json":"https://pith.science/api/pith-number/OGZ2ETE6LULHZ3GKEDKT6COOQU/events.json","paper":"https://pith.science/paper/OGZ2ETE6"},"agent_actions":{"view_html":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU","download_json":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU.json","view_paper":"https://pith.science/paper/OGZ2ETE6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.06516&json=true","fetch_graph":"https://pith.science/api/pith-number/OGZ2ETE6LULHZ3GKEDKT6COOQU/graph.json","fetch_events":"https://pith.science/api/pith-number/OGZ2ETE6LULHZ3GKEDKT6COOQU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU/action/storage_attestation","attest_author":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU/action/author_attestation","sign_citation":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU/action/citation_signature","submit_replication":"https://pith.science/pith/OGZ2ETE6LULHZ3GKEDKT6COOQU/action/replication_record"}},"created_at":"2026-05-20T00:02:58.275734+00:00","updated_at":"2026-05-20T00:02:58.275734+00:00"}