{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZLVRPNYBCFTDQCCL73AO6JDACM","short_pith_number":"pith:ZLVRPNYB","schema_version":"1.0","canonical_sha256":"caeb17b701116638084bfec0ef24601336db9007d24c6311ce4b53da31a2f959","source":{"kind":"arxiv","id":"2510.07257","version":2},"attestation_state":"computed","paper":{"title":"Test-Time Graph Search for Goal-Conditioned Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Claas Voelcker, Evgenii Opryshko, Igor Gilitschenski, Junwei Quan, Yilun Du","submitted_at":"2025-10-08T17:20:53Z","abstract_excerpt":"Offline goal-conditioned reinforcement learning (GCRL) often struggles with long-horizon tasks, where errors in value estimation accumulate and produce unreliable policies. It is typically assumed that effective long-term planning is infeasible without specialized training. In contrast, our work demonstrates that existing GCRL policies can complete long-horizon tasks when combined with a lightweight, training-free planning wrapper. We find that standard goal-conditioned value functions encode locally consistent geometric structure sufficient for planning. Our approach, Test-Time Graph Search ("},"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":"2510.07257","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-10-08T17:20:53Z","cross_cats_sorted":[],"title_canon_sha256":"c8d7003ba192dd56fcca883e13b181dbe54b4c58808734cf04f1a97620d26538","abstract_canon_sha256":"c260e8ae01c100c05648829aa543038a6701b1468e707f0fcf7d08ecbba94d8c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:02:28.239852Z","signature_b64":"sUGSeRb4Kcvm5Dq8uwdd2g+oEym0VbXDjU5tHLReM8ZCTGsf60W32C2FHoNP/iCIjvepI9Dc0NdXihNQw4ZsBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"caeb17b701116638084bfec0ef24601336db9007d24c6311ce4b53da31a2f959","last_reissued_at":"2026-05-26T01:02:28.239038Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:02:28.239038Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Test-Time Graph Search for Goal-Conditioned Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Claas Voelcker, Evgenii Opryshko, Igor Gilitschenski, Junwei Quan, Yilun Du","submitted_at":"2025-10-08T17:20:53Z","abstract_excerpt":"Offline goal-conditioned reinforcement learning (GCRL) often struggles with long-horizon tasks, where errors in value estimation accumulate and produce unreliable policies. It is typically assumed that effective long-term planning is infeasible without specialized training. In contrast, our work demonstrates that existing GCRL policies can complete long-horizon tasks when combined with a lightweight, training-free planning wrapper. We find that standard goal-conditioned value functions encode locally consistent geometric structure sufficient for planning. Our approach, Test-Time Graph Search ("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.07257","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/2510.07257/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":"2510.07257","created_at":"2026-05-26T01:02:28.239156+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.07257v2","created_at":"2026-05-26T01:02:28.239156+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.07257","created_at":"2026-05-26T01:02:28.239156+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZLVRPNYBCFTD","created_at":"2026-05-26T01:02:28.239156+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZLVRPNYBCFTDQCCL","created_at":"2026-05-26T01:02:28.239156+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZLVRPNYB","created_at":"2026-05-26T01:02:28.239156+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2512.00775","citing_title":"SAGAS: Semantic-Aware Graph-Assisted Stitching for Offline Temporal Logic Planning","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2605.03075","citing_title":"Refining Compositional Diffusion for Reliable Long-Horizon Planning","ref_index":54,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM","json":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM.json","graph_json":"https://pith.science/api/pith-number/ZLVRPNYBCFTDQCCL73AO6JDACM/graph.json","events_json":"https://pith.science/api/pith-number/ZLVRPNYBCFTDQCCL73AO6JDACM/events.json","paper":"https://pith.science/paper/ZLVRPNYB"},"agent_actions":{"view_html":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM","download_json":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM.json","view_paper":"https://pith.science/paper/ZLVRPNYB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.07257&json=true","fetch_graph":"https://pith.science/api/pith-number/ZLVRPNYBCFTDQCCL73AO6JDACM/graph.json","fetch_events":"https://pith.science/api/pith-number/ZLVRPNYBCFTDQCCL73AO6JDACM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM/action/storage_attestation","attest_author":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM/action/author_attestation","sign_citation":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM/action/citation_signature","submit_replication":"https://pith.science/pith/ZLVRPNYBCFTDQCCL73AO6JDACM/action/replication_record"}},"created_at":"2026-05-26T01:02:28.239156+00:00","updated_at":"2026-05-26T01:02:28.239156+00:00"}