{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:U2J62HPXZFLJWHMVCTKJAVC5MQ","short_pith_number":"pith:U2J62HPX","schema_version":"1.0","canonical_sha256":"a693ed1df7c9569b1d9514d490545d642423400c660385dbccecc2c5430d71cd","source":{"kind":"arxiv","id":"2505.17138","version":5},"attestation_state":"computed","paper":{"title":"RAP: Runtime Adaptive Pruning for LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chunlin Tian, Huanrong Liu, Li Li, Qingbiao Li, Xuyang Wei","submitted_at":"2025-05-22T06:12:42Z","abstract_excerpt":"Large language models (LLMs) excel at language understanding and generation, but their enormous computational and memory requirements hinder deployment. Compression offers a potential solution to mitigate these constraints. However, most existing methods rely on fixed heuristics and thus fail to adapt to runtime memory variations or heterogeneous KV-cache demands arising from diverse user requests. To address these limitations, we propose RAP, an elastic pruning framework driven by reinforcement learning (RL) that dynamically adjusts compression strategies in a runtime-aware manner. Specifical"},"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":"2505.17138","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-22T06:12:42Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cf26983467c27f9e3f73519939c98bd9646e9df7f6852bc23da60e99f91a0e3d","abstract_canon_sha256":"bc76938da39f21ac8949eb86f3f62191f943f5508e23356bc20616fefeb65489"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:05:29.059080Z","signature_b64":"tyFeSOhD8SmfQzPTaigxZMNFd3FWun2Qxcz7VOLkS+yzEyUZI3oWZj76Yv9PLnTel0kodRqvw87mtuFq9+HpAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a693ed1df7c9569b1d9514d490545d642423400c660385dbccecc2c5430d71cd","last_reissued_at":"2026-05-20T00:05:29.058178Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:05:29.058178Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RAP: Runtime Adaptive Pruning for LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chunlin Tian, Huanrong Liu, Li Li, Qingbiao Li, Xuyang Wei","submitted_at":"2025-05-22T06:12:42Z","abstract_excerpt":"Large language models (LLMs) excel at language understanding and generation, but their enormous computational and memory requirements hinder deployment. Compression offers a potential solution to mitigate these constraints. However, most existing methods rely on fixed heuristics and thus fail to adapt to runtime memory variations or heterogeneous KV-cache demands arising from diverse user requests. To address these limitations, we propose RAP, an elastic pruning framework driven by reinforcement learning (RL) that dynamically adjusts compression strategies in a runtime-aware manner. Specifical"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17138","kind":"arxiv","version":5},"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/2505.17138/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":"2505.17138","created_at":"2026-05-20T00:05:29.058340+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.17138v5","created_at":"2026-05-20T00:05:29.058340+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17138","created_at":"2026-05-20T00:05:29.058340+00:00"},{"alias_kind":"pith_short_12","alias_value":"U2J62HPXZFLJ","created_at":"2026-05-20T00:05:29.058340+00:00"},{"alias_kind":"pith_short_16","alias_value":"U2J62HPXZFLJWHMV","created_at":"2026-05-20T00:05:29.058340+00:00"},{"alias_kind":"pith_short_8","alias_value":"U2J62HPX","created_at":"2026-05-20T00:05:29.058340+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/U2J62HPXZFLJWHMVCTKJAVC5MQ","json":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ.json","graph_json":"https://pith.science/api/pith-number/U2J62HPXZFLJWHMVCTKJAVC5MQ/graph.json","events_json":"https://pith.science/api/pith-number/U2J62HPXZFLJWHMVCTKJAVC5MQ/events.json","paper":"https://pith.science/paper/U2J62HPX"},"agent_actions":{"view_html":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ","download_json":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ.json","view_paper":"https://pith.science/paper/U2J62HPX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.17138&json=true","fetch_graph":"https://pith.science/api/pith-number/U2J62HPXZFLJWHMVCTKJAVC5MQ/graph.json","fetch_events":"https://pith.science/api/pith-number/U2J62HPXZFLJWHMVCTKJAVC5MQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ/action/storage_attestation","attest_author":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ/action/author_attestation","sign_citation":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ/action/citation_signature","submit_replication":"https://pith.science/pith/U2J62HPXZFLJWHMVCTKJAVC5MQ/action/replication_record"}},"created_at":"2026-05-20T00:05:29.058340+00:00","updated_at":"2026-05-20T00:05:29.058340+00:00"}