{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IZORAA2TKSCMWDLH3SVMNLPKNF","short_pith_number":"pith:IZORAA2T","schema_version":"1.0","canonical_sha256":"465d1003535484cb0d67dcaac6adea6960c232b65e9f06ee3c3ae489d81b45a4","source":{"kind":"arxiv","id":"2606.02132","version":1},"attestation_state":"computed","paper":{"title":"Learning When Not to Act: Mitigating Tool Abuse in Agentic Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dianxing Tang, Dingshuo Chen, Liang Wang, Liuji Chen, Qiang Liu, Shu Wu, Xing Shi","submitted_at":"2026-06-01T11:58:55Z","abstract_excerpt":"Agentic reinforcement learning can induce tool abuse, where models overuse external tools even for queries solvable by internal reasoning. Existing approaches mitigate this issue with uniform tool-use penalties or hard limits, which reduce tool frequency but may also suppress useful tool-assisted exploration. We propose EAPO, an Efficient Agentic Policy Optimization framework that learns selective tool use. EAPO introduces tool-free trajectories into each rollout group, applies difficulty-aware reward shaping to penalize redundant tool calls mainly on easier queries, and uses confidence-aware "},"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":"2606.02132","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-01T11:58:55Z","cross_cats_sorted":[],"title_canon_sha256":"d9964c827ba10a7d3a04452ff9bf28ee36a83f8bc38c59ca3833d18fdef1b550","abstract_canon_sha256":"9f5472d381d7dae585da1eb8dd911a758f6cc6d72267d5dc9bd08e53a0883c30"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:05:07.340858Z","signature_b64":"dRKNW4MnSjQNGEkug+dSM0yud9qpAWusWsSiC3x2MK4AQnQ6bmNgzNeNG9RS+xpv9GMjbeuNje3egDqMg5cYBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"465d1003535484cb0d67dcaac6adea6960c232b65e9f06ee3c3ae489d81b45a4","last_reissued_at":"2026-06-02T02:05:07.340464Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:05:07.340464Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning When Not to Act: Mitigating Tool Abuse in Agentic Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Dianxing Tang, Dingshuo Chen, Liang Wang, Liuji Chen, Qiang Liu, Shu Wu, Xing Shi","submitted_at":"2026-06-01T11:58:55Z","abstract_excerpt":"Agentic reinforcement learning can induce tool abuse, where models overuse external tools even for queries solvable by internal reasoning. Existing approaches mitigate this issue with uniform tool-use penalties or hard limits, which reduce tool frequency but may also suppress useful tool-assisted exploration. We propose EAPO, an Efficient Agentic Policy Optimization framework that learns selective tool use. EAPO introduces tool-free trajectories into each rollout group, applies difficulty-aware reward shaping to penalize redundant tool calls mainly on easier queries, and uses confidence-aware "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.02132","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/2606.02132/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":"2606.02132","created_at":"2026-06-02T02:05:07.340521+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.02132v1","created_at":"2026-06-02T02:05:07.340521+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.02132","created_at":"2026-06-02T02:05:07.340521+00:00"},{"alias_kind":"pith_short_12","alias_value":"IZORAA2TKSCM","created_at":"2026-06-02T02:05:07.340521+00:00"},{"alias_kind":"pith_short_16","alias_value":"IZORAA2TKSCMWDLH","created_at":"2026-06-02T02:05:07.340521+00:00"},{"alias_kind":"pith_short_8","alias_value":"IZORAA2T","created_at":"2026-06-02T02:05:07.340521+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/IZORAA2TKSCMWDLH3SVMNLPKNF","json":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF.json","graph_json":"https://pith.science/api/pith-number/IZORAA2TKSCMWDLH3SVMNLPKNF/graph.json","events_json":"https://pith.science/api/pith-number/IZORAA2TKSCMWDLH3SVMNLPKNF/events.json","paper":"https://pith.science/paper/IZORAA2T"},"agent_actions":{"view_html":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF","download_json":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF.json","view_paper":"https://pith.science/paper/IZORAA2T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.02132&json=true","fetch_graph":"https://pith.science/api/pith-number/IZORAA2TKSCMWDLH3SVMNLPKNF/graph.json","fetch_events":"https://pith.science/api/pith-number/IZORAA2TKSCMWDLH3SVMNLPKNF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF/action/storage_attestation","attest_author":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF/action/author_attestation","sign_citation":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF/action/citation_signature","submit_replication":"https://pith.science/pith/IZORAA2TKSCMWDLH3SVMNLPKNF/action/replication_record"}},"created_at":"2026-06-02T02:05:07.340521+00:00","updated_at":"2026-06-02T02:05:07.340521+00:00"}