{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:62HQLE7D4GUQK4DAUGOWDC2VMG","short_pith_number":"pith:62HQLE7D","schema_version":"1.0","canonical_sha256":"f68f0593e3e1a9057060a19d618b55618e19ac89457c928cbb9d156793f378e2","source":{"kind":"arxiv","id":"2606.17524","version":1},"attestation_state":"computed","paper":{"title":"Learning to Refine Hidden States for Reliable LLM Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chia-Hsuan Hsu, Jui-Ming Yao","submitted_at":"2026-06-16T05:03:27Z","abstract_excerpt":"Large language models show strong reasoning ability, but their internal reasoning process can remain unstable in complex multi-step settings, where early hidden-state errors may propagate to incorrect predictions. We propose ReLAR, a reinforcement-guided latent refinement framework that iteratively updates hidden representations before decoding. ReLAR maintains a compact latent reasoning state and uses learned depth and action controllers to adaptively determine both the number and direction of refinement steps. The controllers are trained with a policy gradient objective based on step-wise li"},"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":"2606.17524","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-16T05:03:27Z","cross_cats_sorted":[],"title_canon_sha256":"4b6a880a8b5bac3c4916fdf0353f70063f82de373e2d59e4136152c8ae56468f","abstract_canon_sha256":"0b54dac172dabe855c23e785df4d5ad5ddfa0615e6f8f1d7beda6a21cc19c393"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:10:15.103445Z","signature_b64":"m3AZbqr8XrDMKOMb46UBAQNBnQGdE87o7q9VEE814hacOlbd3cgYMIBttoAn+gjLwwvAGDe89KhGM0pg2viiAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f68f0593e3e1a9057060a19d618b55618e19ac89457c928cbb9d156793f378e2","last_reissued_at":"2026-06-19T16:10:15.103081Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:10:15.103081Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning to Refine Hidden States for Reliable LLM Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chia-Hsuan Hsu, Jui-Ming Yao","submitted_at":"2026-06-16T05:03:27Z","abstract_excerpt":"Large language models show strong reasoning ability, but their internal reasoning process can remain unstable in complex multi-step settings, where early hidden-state errors may propagate to incorrect predictions. We propose ReLAR, a reinforcement-guided latent refinement framework that iteratively updates hidden representations before decoding. ReLAR maintains a compact latent reasoning state and uses learned depth and action controllers to adaptively determine both the number and direction of refinement steps. The controllers are trained with a policy gradient objective based on step-wise li"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.17524","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/2606.17524/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":"2606.17524","created_at":"2026-06-19T16:10:15.103143+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.17524v1","created_at":"2026-06-19T16:10:15.103143+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.17524","created_at":"2026-06-19T16:10:15.103143+00:00"},{"alias_kind":"pith_short_12","alias_value":"62HQLE7D4GUQ","created_at":"2026-06-19T16:10:15.103143+00:00"},{"alias_kind":"pith_short_16","alias_value":"62HQLE7D4GUQK4DA","created_at":"2026-06-19T16:10:15.103143+00:00"},{"alias_kind":"pith_short_8","alias_value":"62HQLE7D","created_at":"2026-06-19T16:10:15.103143+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/62HQLE7D4GUQK4DAUGOWDC2VMG","json":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG.json","graph_json":"https://pith.science/api/pith-number/62HQLE7D4GUQK4DAUGOWDC2VMG/graph.json","events_json":"https://pith.science/api/pith-number/62HQLE7D4GUQK4DAUGOWDC2VMG/events.json","paper":"https://pith.science/paper/62HQLE7D"},"agent_actions":{"view_html":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG","download_json":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG.json","view_paper":"https://pith.science/paper/62HQLE7D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.17524&json=true","fetch_graph":"https://pith.science/api/pith-number/62HQLE7D4GUQK4DAUGOWDC2VMG/graph.json","fetch_events":"https://pith.science/api/pith-number/62HQLE7D4GUQK4DAUGOWDC2VMG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG/action/storage_attestation","attest_author":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG/action/author_attestation","sign_citation":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG/action/citation_signature","submit_replication":"https://pith.science/pith/62HQLE7D4GUQK4DAUGOWDC2VMG/action/replication_record"}},"created_at":"2026-06-19T16:10:15.103143+00:00","updated_at":"2026-06-19T16:10:15.103143+00:00"}