{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:TKKPQ7VGF4FQ6KF3PIGUYON6Z5","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"e9e37f6ebf66ac0ba1c462b7cca11bf0e1202c7332a55398885637bfdbe9e707","cross_cats_sorted":["cs.AI","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-24T01:31:05Z","title_canon_sha256":"a7852d851c0de32cb089099cbc0db4cd2851155bc42aa0ecb666c25d95f2b325"},"schema_version":"1.0","source":{"id":"2504.18587","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.18587","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"arxiv_version","alias_value":"2504.18587v1","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.18587","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_12","alias_value":"TKKPQ7VGF4FQ","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_16","alias_value":"TKKPQ7VGF4FQ6KF3","created_at":"2026-07-05T10:54:17Z"},{"alias_kind":"pith_short_8","alias_value":"TKKPQ7VG","created_at":"2026-07-05T10:54:17Z"}],"graph_snapshots":[{"event_id":"sha256:59815b52b9927b28c3e9ebb945cc64ea92cc81feb109bfa06a202366e021b0d2","target":"graph","created_at":"2026-07-05T10:54:17Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2504.18587/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Recently, foundation models such as OpenAI's O1 and O3, along with DeepSeek's R1, have demonstrated strong reasoning capacities and problem-solving skills acquired through large-scale reinforcement learning (RL), with wide applications in mathematics, coding, science, intelligent agents, and virtual assistants. In this work, we introduce an off-policy reinforcement learning algorithm, EM Policy Gradient, aimed at enhancing LLM reasoning by optimizing expected return over reasoning trajectories. We frame the reasoning task as an Expectation-Maximization (EM) optimization problem, alternating be","authors_text":"Tianbing Xu","cross_cats":["cs.AI","stat.ML"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-24T01:31:05Z","title":"Training Large Language Models to Reason via EM Policy Gradient"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.18587","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ca588f860f8585ba651ef68e1c3b60e8ca4c2a33f027347d6587c9e73ce0667f","target":"record","created_at":"2026-07-05T10:54:17Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"e9e37f6ebf66ac0ba1c462b7cca11bf0e1202c7332a55398885637bfdbe9e707","cross_cats_sorted":["cs.AI","stat.ML"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-24T01:31:05Z","title_canon_sha256":"a7852d851c0de32cb089099cbc0db4cd2851155bc42aa0ecb666c25d95f2b325"},"schema_version":"1.0","source":{"id":"2504.18587","kind":"arxiv","version":1}},"canonical_sha256":"9a94f87ea62f0b0f28bb7a0d4c39becf6512157f0fa73352978c436b1695acd6","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9a94f87ea62f0b0f28bb7a0d4c39becf6512157f0fa73352978c436b1695acd6","first_computed_at":"2026-07-05T10:54:17.501845Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:54:17.501845Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"oSSUvOOuCRnrLDo3BpyCv3cjnK22aVkVK55K+oJwnT0pKOnESlckicoxYQkf+ZAzsSBsKe+fom/cYUlGwx7yBg==","signature_status":"signed_v1","signed_at":"2026-07-05T10:54:17.502334Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.18587","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ca588f860f8585ba651ef68e1c3b60e8ca4c2a33f027347d6587c9e73ce0667f","sha256:59815b52b9927b28c3e9ebb945cc64ea92cc81feb109bfa06a202366e021b0d2"],"state_sha256":"7ba13f0f31123d26697a75dbc648310fcabf55c878b00521b57f03d6f05cc55c"}