{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HBE2GPYFAJ5XDJVIT4XKA64ADC","short_pith_number":"pith:HBE2GPYF","schema_version":"1.0","canonical_sha256":"3849a33f05027b71a6a89f2ea07b8018ba38cfeaec0f8eb6342cdf6d3e96cca8","source":{"kind":"arxiv","id":"2505.17391","version":1},"attestation_state":"computed","paper":{"title":"Curriculum Guided Reinforcement Learning for Efficient Multi Hop Retrieval Augmented Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Daqing He, Rui Meng, Yuelyu Ji, Zhuochun Li","submitted_at":"2025-05-23T02:01:15Z","abstract_excerpt":"Retrieval-augmented generation (RAG) grounds large language models (LLMs) in up-to-date external evidence, yet existing multi-hop RAG pipelines still issue redundant subqueries, explore too shallowly, or wander through overly long search chains. We introduce EVO-RAG, a curriculum-guided reinforcement learning framework that evolves a query-rewriting agent from broad early-stage exploration to concise late-stage refinement. EVO-RAG couples a seven-factor, step-level reward vector (covering relevance, redundancy, efficiency, and answer correctness) with a time-varying scheduler that reweights th"},"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":"2505.17391","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-23T02:01:15Z","cross_cats_sorted":[],"title_canon_sha256":"6435c1121fb4b93d0fe24e518ae3e33839f2ed0d79bcfd651980d2f4da03f363","abstract_canon_sha256":"52b20b78c44f63c2984a34da15c2267cda215fe2224dfcf14a470a72fd127131"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:08:25.256452Z","signature_b64":"NykGJLft+WjUlpXX7MfNFIe+FR7WIfnHjiHlg8mDvCqvFRNz0htFL+JBaSX/LD55E13HdzI+47jYW8p6L21tDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3849a33f05027b71a6a89f2ea07b8018ba38cfeaec0f8eb6342cdf6d3e96cca8","last_reissued_at":"2026-07-05T11:08:25.255902Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:08:25.255902Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Curriculum Guided Reinforcement Learning for Efficient Multi Hop Retrieval Augmented Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Daqing He, Rui Meng, Yuelyu Ji, Zhuochun Li","submitted_at":"2025-05-23T02:01:15Z","abstract_excerpt":"Retrieval-augmented generation (RAG) grounds large language models (LLMs) in up-to-date external evidence, yet existing multi-hop RAG pipelines still issue redundant subqueries, explore too shallowly, or wander through overly long search chains. We introduce EVO-RAG, a curriculum-guided reinforcement learning framework that evolves a query-rewriting agent from broad early-stage exploration to concise late-stage refinement. EVO-RAG couples a seven-factor, step-level reward vector (covering relevance, redundancy, efficiency, and answer correctness) with a time-varying scheduler that reweights th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17391","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/2505.17391/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":"2505.17391","created_at":"2026-07-05T11:08:25.255964+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.17391v1","created_at":"2026-07-05T11:08:25.255964+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17391","created_at":"2026-07-05T11:08:25.255964+00:00"},{"alias_kind":"pith_short_12","alias_value":"HBE2GPYFAJ5X","created_at":"2026-07-05T11:08:25.255964+00:00"},{"alias_kind":"pith_short_16","alias_value":"HBE2GPYFAJ5XDJVI","created_at":"2026-07-05T11:08:25.255964+00:00"},{"alias_kind":"pith_short_8","alias_value":"HBE2GPYF","created_at":"2026-07-05T11:08:25.255964+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.24660","citing_title":"How Many Tools Should an LLM Agent See? A Chance-Corrected Answer","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC","json":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC.json","graph_json":"https://pith.science/api/pith-number/HBE2GPYFAJ5XDJVIT4XKA64ADC/graph.json","events_json":"https://pith.science/api/pith-number/HBE2GPYFAJ5XDJVIT4XKA64ADC/events.json","paper":"https://pith.science/paper/HBE2GPYF"},"agent_actions":{"view_html":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC","download_json":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC.json","view_paper":"https://pith.science/paper/HBE2GPYF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.17391&json=true","fetch_graph":"https://pith.science/api/pith-number/HBE2GPYFAJ5XDJVIT4XKA64ADC/graph.json","fetch_events":"https://pith.science/api/pith-number/HBE2GPYFAJ5XDJVIT4XKA64ADC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC/action/storage_attestation","attest_author":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC/action/author_attestation","sign_citation":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC/action/citation_signature","submit_replication":"https://pith.science/pith/HBE2GPYFAJ5XDJVIT4XKA64ADC/action/replication_record"}},"created_at":"2026-07-05T11:08:25.255964+00:00","updated_at":"2026-07-05T11:08:25.255964+00:00"}