{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IRIPJL66M5GEYBBDQ7GN6KYRFO","short_pith_number":"pith:IRIPJL66","schema_version":"1.0","canonical_sha256":"4450f4afde674c4c042387ccdf2b112b9369b198c442afa0dca2ac47442ef213","source":{"kind":"arxiv","id":"2507.07451","version":1},"attestation_state":"computed","paper":{"title":"RLEP: Reinforcement Learning with Experience Replay for LLM Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Fuzheng Zhang, Guorui Zhou, Hongzhi Zhang, Jia Fu, Jingyuan Zhang, Kai Fu, Qi Wang","submitted_at":"2025-07-10T05:58:55Z","abstract_excerpt":"Reinforcement learning (RL) for large language models is an energy-intensive endeavor: training can be unstable, and the policy may gradually drift away from its pretrained weights. We present \\emph{RLEP}\\, -- \\,Reinforcement Learning with Experience rePlay\\, -- \\,a two-phase framework that first collects verified trajectories and then replays them during subsequent training. At every update step, the policy is optimized on mini-batches that blend newly generated rollouts with these replayed successes. By replaying high-quality examples, RLEP steers the model away from fruitless exploration, f"},"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":"2507.07451","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-10T05:58:55Z","cross_cats_sorted":[],"title_canon_sha256":"6e940d0f99b315f7f13da3fcef0f3f41da94f264a9def7689ae5734cc39d768b","abstract_canon_sha256":"69abf5140e9a5657daa7d2b3fee2aec8f3a1292dab5b71fb107c28e6321bdd56"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:34:58.442965Z","signature_b64":"ChrphzzlKj8Q8Lx8GcW5N5ZAlwQM5q1oRhw2C1/UKPczvf8Lb4EW9je2tS19gbaLkyw8B+boZJmkYNZdd3JHDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4450f4afde674c4c042387ccdf2b112b9369b198c442afa0dca2ac47442ef213","last_reissued_at":"2026-07-05T11:34:58.442516Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:34:58.442516Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RLEP: Reinforcement Learning with Experience Replay for LLM Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Fuzheng Zhang, Guorui Zhou, Hongzhi Zhang, Jia Fu, Jingyuan Zhang, Kai Fu, Qi Wang","submitted_at":"2025-07-10T05:58:55Z","abstract_excerpt":"Reinforcement learning (RL) for large language models is an energy-intensive endeavor: training can be unstable, and the policy may gradually drift away from its pretrained weights. We present \\emph{RLEP}\\, -- \\,Reinforcement Learning with Experience rePlay\\, -- \\,a two-phase framework that first collects verified trajectories and then replays them during subsequent training. At every update step, the policy is optimized on mini-batches that blend newly generated rollouts with these replayed successes. By replaying high-quality examples, RLEP steers the model away from fruitless exploration, f"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.07451","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/2507.07451/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":"2507.07451","created_at":"2026-07-05T11:34:58.442575+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.07451v1","created_at":"2026-07-05T11:34:58.442575+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.07451","created_at":"2026-07-05T11:34:58.442575+00:00"},{"alias_kind":"pith_short_12","alias_value":"IRIPJL66M5GE","created_at":"2026-07-05T11:34:58.442575+00:00"},{"alias_kind":"pith_short_16","alias_value":"IRIPJL66M5GEYBBD","created_at":"2026-07-05T11:34:58.442575+00:00"},{"alias_kind":"pith_short_8","alias_value":"IRIPJL66","created_at":"2026-07-05T11:34:58.442575+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21943","citing_title":"Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning","ref_index":261,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18216","citing_title":"Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04560","citing_title":"Rollout-Level Advantage-Prioritized Experience Replay for GRPO","ref_index":76,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03087","citing_title":"Learning to Solve, Forgetting to Retain: Correct-Set Turnover in RLVR","ref_index":143,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01281","citing_title":"RLVR without Ineffective Samples: Group Prioritized Off-Policy Optimization for LLM Reasoning","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01249","citing_title":"Trust Region On-Policy Distillation","ref_index":275,"is_internal_anchor":false},{"citing_arxiv_id":"2606.30345","citing_title":"DRIFT: Difficulty Routing Self-DIstillation with Rhythm-Gated Exploration and Success BuFfer Training","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.30420","citing_title":"Experience Augmented Policy Optimization for LLM Reasoning","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07331","citing_title":"Rethinking Importance Sampling in LLM Policy Optimization: A Cumulative Token Perspective","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16893","citing_title":"EasyVideoR1: Easier RL for Video Understanding","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20733","citing_title":"Near-Future Policy Optimization","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO","json":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO.json","graph_json":"https://pith.science/api/pith-number/IRIPJL66M5GEYBBDQ7GN6KYRFO/graph.json","events_json":"https://pith.science/api/pith-number/IRIPJL66M5GEYBBDQ7GN6KYRFO/events.json","paper":"https://pith.science/paper/IRIPJL66"},"agent_actions":{"view_html":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO","download_json":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO.json","view_paper":"https://pith.science/paper/IRIPJL66","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.07451&json=true","fetch_graph":"https://pith.science/api/pith-number/IRIPJL66M5GEYBBDQ7GN6KYRFO/graph.json","fetch_events":"https://pith.science/api/pith-number/IRIPJL66M5GEYBBDQ7GN6KYRFO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO/action/storage_attestation","attest_author":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO/action/author_attestation","sign_citation":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO/action/citation_signature","submit_replication":"https://pith.science/pith/IRIPJL66M5GEYBBDQ7GN6KYRFO/action/replication_record"}},"created_at":"2026-07-05T11:34:58.442575+00:00","updated_at":"2026-07-05T11:34:58.442575+00:00"}