{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UT2HHMIYGRXPO4YLFJ6VDQ3AUJ","short_pith_number":"pith:UT2HHMIY","schema_version":"1.0","canonical_sha256":"a4f473b118346ef7730b2a7d51c360a25a9fbe6c9b43af9f115ee42b115f3df0","source":{"kind":"arxiv","id":"2503.07518","version":2},"attestation_state":"computed","paper":{"title":"TokenButler: Token Importance is Predictable","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ahmed F AbouElhamayed, Chi-Chih Chang, Mohamed S. Abdelfattah, Nilesh Jain, Sameh Gobriel, Yash Akhauri, Yifei Gao","submitted_at":"2025-03-10T16:41:14Z","abstract_excerpt":"Large Language Models (LLMs) rely on the Key-Value (KV) Cache to store token history, enabling efficient decoding of tokens. As the KV-Cache grows, it becomes a major memory and computation bottleneck. However, there is an opportunity to alleviate this bottleneck, prior research has shown that only a small subset of tokens contribute meaningfully to each decoding step. A key challenge in finding these critical tokens is that they are dynamic, and heavily input query-dependent. Existing methods either risk quality by evicting tokens permanently, or retain the full KV-Cache but rely on retrievin"},"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":"2503.07518","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-10T16:41:14Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"2de7500fe025917d761a7d514ad9f97dd447e4ba5f872f585f435fff9dfd7261","abstract_canon_sha256":"28dde252ec302a2dc2767147220ed44531be318e4601e56ec93e6fd4a984d621"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:18.104053Z","signature_b64":"7dnlLGSoVSOXJkMZM26Mi+EvDkLRNgTnNRRgTfD+fIP9QCTMJBZqeLMMyi1fwMwf97YC+La8cueKq2jxQoB9DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a4f473b118346ef7730b2a7d51c360a25a9fbe6c9b43af9f115ee42b115f3df0","last_reissued_at":"2026-05-20T00:00:18.103381Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:18.103381Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TokenButler: Token Importance is Predictable","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ahmed F AbouElhamayed, Chi-Chih Chang, Mohamed S. Abdelfattah, Nilesh Jain, Sameh Gobriel, Yash Akhauri, Yifei Gao","submitted_at":"2025-03-10T16:41:14Z","abstract_excerpt":"Large Language Models (LLMs) rely on the Key-Value (KV) Cache to store token history, enabling efficient decoding of tokens. As the KV-Cache grows, it becomes a major memory and computation bottleneck. However, there is an opportunity to alleviate this bottleneck, prior research has shown that only a small subset of tokens contribute meaningfully to each decoding step. A key challenge in finding these critical tokens is that they are dynamic, and heavily input query-dependent. Existing methods either risk quality by evicting tokens permanently, or retain the full KV-Cache but rely on retrievin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.07518","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/2503.07518/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":"2503.07518","created_at":"2026-05-20T00:00:18.103522+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.07518v2","created_at":"2026-05-20T00:00:18.103522+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.07518","created_at":"2026-05-20T00:00:18.103522+00:00"},{"alias_kind":"pith_short_12","alias_value":"UT2HHMIYGRXP","created_at":"2026-05-20T00:00:18.103522+00:00"},{"alias_kind":"pith_short_16","alias_value":"UT2HHMIYGRXPO4YL","created_at":"2026-05-20T00:00:18.103522+00:00"},{"alias_kind":"pith_short_8","alias_value":"UT2HHMIY","created_at":"2026-05-20T00:00:18.103522+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2605.17410","citing_title":"Computational Challenges in Token Economics: Bridging Economic Theory and AI System Design","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10875","citing_title":"Compute Where it Counts: Self Optimizing Language Models","ref_index":2,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ","json":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ.json","graph_json":"https://pith.science/api/pith-number/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/graph.json","events_json":"https://pith.science/api/pith-number/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/events.json","paper":"https://pith.science/paper/UT2HHMIY"},"agent_actions":{"view_html":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ","download_json":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ.json","view_paper":"https://pith.science/paper/UT2HHMIY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.07518&json=true","fetch_graph":"https://pith.science/api/pith-number/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/graph.json","fetch_events":"https://pith.science/api/pith-number/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/action/storage_attestation","attest_author":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/action/author_attestation","sign_citation":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/action/citation_signature","submit_replication":"https://pith.science/pith/UT2HHMIYGRXPO4YLFJ6VDQ3AUJ/action/replication_record"}},"created_at":"2026-05-20T00:00:18.103522+00:00","updated_at":"2026-05-20T00:00:18.103522+00:00"}