{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:WF3DS6WHGQB47OYLDOZ6QOIKRU","short_pith_number":"pith:WF3DS6WH","schema_version":"1.0","canonical_sha256":"b176397ac73403cfbb0b1bb3e8390a8d1fea44afdc7043100908940d4a16c61c","source":{"kind":"arxiv","id":"2606.25852","version":1},"attestation_state":"computed","paper":{"title":"Semantic Consistency Policy Optimization for Reinforcement Learning of LLM Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Junzhuo Li, Peng Xu, Sijia Chen, Xuming Hu","submitted_at":"2026-06-24T14:02:13Z","abstract_excerpt":"Group-based reinforcement learning effectively post-trains LLM agents for long-horizon, sparse-reward tasks by deriving step-level credit from trajectory outcomes. However, this ties a step's credit to its rollout's final outcome: semantically near-identical intermediate steps receive opposite credit depending on whether their trajectory eventually succeeded or failed. Such semantic credit inconsistency sends conflicting gradients to similar actions and wastes the partially-correct progress inside failed rollouts. Motivated by this, we propose Semantic Consistency Policy Optimization (SCPO), a"},"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":"2606.25852","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-24T14:02:13Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"f406dc07c5cc8cc1893ffb8a67134db49bf6695c2d870c070874179bbde40e63","abstract_canon_sha256":"1dfe2806e5d619c201db4592db771803405fcb705dfc7591b6b29546a0b3ec3d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-25T01:18:41.109463Z","signature_b64":"Y0WpvHhuM5GJYvMnEOlsdifQMQRNGjPL5ZhabwkihujEq6VjYoWoZALr5qTliYRVLVSsjigw0gUh+ZtJbtYMAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b176397ac73403cfbb0b1bb3e8390a8d1fea44afdc7043100908940d4a16c61c","last_reissued_at":"2026-06-25T01:18:41.109079Z","signature_status":"signed_v1","first_computed_at":"2026-06-25T01:18:41.109079Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Semantic Consistency Policy Optimization for Reinforcement Learning of LLM Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Junzhuo Li, Peng Xu, Sijia Chen, Xuming Hu","submitted_at":"2026-06-24T14:02:13Z","abstract_excerpt":"Group-based reinforcement learning effectively post-trains LLM agents for long-horizon, sparse-reward tasks by deriving step-level credit from trajectory outcomes. However, this ties a step's credit to its rollout's final outcome: semantically near-identical intermediate steps receive opposite credit depending on whether their trajectory eventually succeeded or failed. Such semantic credit inconsistency sends conflicting gradients to similar actions and wastes the partially-correct progress inside failed rollouts. Motivated by this, we propose Semantic Consistency Policy Optimization (SCPO), a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.25852","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/2606.25852/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":"2606.25852","created_at":"2026-06-25T01:18:41.109142+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.25852v1","created_at":"2026-06-25T01:18:41.109142+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.25852","created_at":"2026-06-25T01:18:41.109142+00:00"},{"alias_kind":"pith_short_12","alias_value":"WF3DS6WHGQB4","created_at":"2026-06-25T01:18:41.109142+00:00"},{"alias_kind":"pith_short_16","alias_value":"WF3DS6WHGQB47OYL","created_at":"2026-06-25T01:18:41.109142+00:00"},{"alias_kind":"pith_short_8","alias_value":"WF3DS6WH","created_at":"2026-06-25T01:18:41.109142+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/WF3DS6WHGQB47OYLDOZ6QOIKRU","json":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU.json","graph_json":"https://pith.science/api/pith-number/WF3DS6WHGQB47OYLDOZ6QOIKRU/graph.json","events_json":"https://pith.science/api/pith-number/WF3DS6WHGQB47OYLDOZ6QOIKRU/events.json","paper":"https://pith.science/paper/WF3DS6WH"},"agent_actions":{"view_html":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU","download_json":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU.json","view_paper":"https://pith.science/paper/WF3DS6WH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.25852&json=true","fetch_graph":"https://pith.science/api/pith-number/WF3DS6WHGQB47OYLDOZ6QOIKRU/graph.json","fetch_events":"https://pith.science/api/pith-number/WF3DS6WHGQB47OYLDOZ6QOIKRU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU/action/storage_attestation","attest_author":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU/action/author_attestation","sign_citation":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU/action/citation_signature","submit_replication":"https://pith.science/pith/WF3DS6WHGQB47OYLDOZ6QOIKRU/action/replication_record"}},"created_at":"2026-06-25T01:18:41.109142+00:00","updated_at":"2026-06-25T01:18:41.109142+00:00"}