{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FW5BW4ZRGLNSZSKOHX4GTTC3ZW","short_pith_number":"pith:FW5BW4ZR","schema_version":"1.0","canonical_sha256":"2dba1b733132db2cc94e3df869cc5bcda99eec09a05e61c34337e11d5656a16e","source":{"kind":"arxiv","id":"2504.01281","version":3},"attestation_state":"computed","paper":{"title":"Scaling Test-Time Inference with Policy-Optimized, Dynamic Retrieval-Augmented Generation via KV Caching and Decoding","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IR"],"primary_cat":"cs.LG","authors_text":"Akash Das, Sakhinana Sagar Srinivas, Shivam Gupta, Venkataramana Runkana","submitted_at":"2025-04-02T01:16:10Z","abstract_excerpt":"We present a comprehensive framework for enhancing Retrieval-Augmented Generation (RAG) systems through dynamic retrieval strategies and reinforcement fine-tuning. This approach significantly improves large language models on knowledge-intensive tasks, including opendomain question answering and complex reasoning. Our framework integrates two complementary techniques: Policy-Optimized RetrievalAugmented Generation (PORAG), which optimizes the use of retrieved information, and Adaptive Token-Layer Attention Scoring (ATLAS), which dynamically determines retrieval timing and content based on cont"},"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":"2504.01281","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-02T01:16:10Z","cross_cats_sorted":["cs.AI","cs.CL","cs.IR"],"title_canon_sha256":"ab5b40f9810910c40aadef04d9a23be81f31dbfe2c81bc44f4450a77be8315fd","abstract_canon_sha256":"f48c1169fe9a6df5c84343f6dccb8744479ac2eff90678e3aeff808c8b676177"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:05:42.812128Z","signature_b64":"0L3+6Dw4mF3usAatgJAmYjmuD0FBM8AKcWvVtSHqd9WkhTvjIac5qBlFgoJjkuszDuph3Q/Pqlf1ZwvBA+N3AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2dba1b733132db2cc94e3df869cc5bcda99eec09a05e61c34337e11d5656a16e","last_reissued_at":"2026-07-05T11:05:42.811518Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:05:42.811518Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Scaling Test-Time Inference with Policy-Optimized, Dynamic Retrieval-Augmented Generation via KV Caching and Decoding","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IR"],"primary_cat":"cs.LG","authors_text":"Akash Das, Sakhinana Sagar Srinivas, Shivam Gupta, Venkataramana Runkana","submitted_at":"2025-04-02T01:16:10Z","abstract_excerpt":"We present a comprehensive framework for enhancing Retrieval-Augmented Generation (RAG) systems through dynamic retrieval strategies and reinforcement fine-tuning. This approach significantly improves large language models on knowledge-intensive tasks, including opendomain question answering and complex reasoning. Our framework integrates two complementary techniques: Policy-Optimized RetrievalAugmented Generation (PORAG), which optimizes the use of retrieved information, and Adaptive Token-Layer Attention Scoring (ATLAS), which dynamically determines retrieval timing and content based on cont"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.01281","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/2504.01281/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":"2504.01281","created_at":"2026-07-05T11:05:42.811579+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.01281v3","created_at":"2026-07-05T11:05:42.811579+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.01281","created_at":"2026-07-05T11:05:42.811579+00:00"},{"alias_kind":"pith_short_12","alias_value":"FW5BW4ZRGLNS","created_at":"2026-07-05T11:05:42.811579+00:00"},{"alias_kind":"pith_short_16","alias_value":"FW5BW4ZRGLNSZSKO","created_at":"2026-07-05T11:05:42.811579+00:00"},{"alias_kind":"pith_short_8","alias_value":"FW5BW4ZR","created_at":"2026-07-05T11:05:42.811579+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.15759","citing_title":"Interaction as Intelligence: Deep Research With Human-AI Partnership","ref_index":82,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW","json":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW.json","graph_json":"https://pith.science/api/pith-number/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/graph.json","events_json":"https://pith.science/api/pith-number/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/events.json","paper":"https://pith.science/paper/FW5BW4ZR"},"agent_actions":{"view_html":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW","download_json":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW.json","view_paper":"https://pith.science/paper/FW5BW4ZR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.01281&json=true","fetch_graph":"https://pith.science/api/pith-number/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/graph.json","fetch_events":"https://pith.science/api/pith-number/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/action/storage_attestation","attest_author":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/action/author_attestation","sign_citation":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/action/citation_signature","submit_replication":"https://pith.science/pith/FW5BW4ZRGLNSZSKOHX4GTTC3ZW/action/replication_record"}},"created_at":"2026-07-05T11:05:42.811579+00:00","updated_at":"2026-07-05T11:05:42.811579+00:00"}