{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MLFKAQMAZE45MHW6FGHQKZ2CCP","short_pith_number":"pith:MLFKAQMA","schema_version":"1.0","canonical_sha256":"62caa04180c939d61ede298f05674213c89b17eddbbf3259ddf8da45da1fcdc7","source":{"kind":"arxiv","id":"2508.07534","version":2},"attestation_state":"computed","paper":{"title":"From Trial-and-Error to Improvement: A Systematic Analysis of LLM Exploration Mechanisms in RLVR","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Beichen Zhang, Daixuan Cheng, Fei Bai, Jia Deng, Jie Chen, Ji-Rong Wen, Wayne Xin Zhao, Yanzipeng Gao, Yinqian Min, Zhipeng Chen","submitted_at":"2025-08-11T01:26:16Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for enhancing the reasoning capabilities of large language models (LLMs). Unlike traditional RL approaches, RLVR leverages rule-based feedback to guide LLMs in generating and refining complex reasoning chains -- a process critically dependent on effective exploration strategies. While prior work has demonstrated RLVR's empirical success, the fundamental mechanisms governing LLMs' exploration behaviors remain underexplored. This technical report presents a systematic investigation of exploration capacities "},"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":"2508.07534","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-08-11T01:26:16Z","cross_cats_sorted":[],"title_canon_sha256":"e24eff13e24dd89753af7249181a7df59f66f032d218cc194459391f4ea7560d","abstract_canon_sha256":"012f9ec60945ee3573c7a94eb5746e600d3ef5fba066dc6d8fe2676fe402e47b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:54:50.774639Z","signature_b64":"etggnVXNy8R5K3Lnfz7HrfaAznSjE7OmftLLLwO9qT4WeIptpKbAEkXXb9b9x8j8hkHFd66jzEc6sJ0ow4VLCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"62caa04180c939d61ede298f05674213c89b17eddbbf3259ddf8da45da1fcdc7","last_reissued_at":"2026-07-05T11:54:50.774164Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:54:50.774164Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"From Trial-and-Error to Improvement: A Systematic Analysis of LLM Exploration Mechanisms in RLVR","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Beichen Zhang, Daixuan Cheng, Fei Bai, Jia Deng, Jie Chen, Ji-Rong Wen, Wayne Xin Zhao, Yanzipeng Gao, Yinqian Min, Zhipeng Chen","submitted_at":"2025-08-11T01:26:16Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has emerged as a powerful paradigm for enhancing the reasoning capabilities of large language models (LLMs). Unlike traditional RL approaches, RLVR leverages rule-based feedback to guide LLMs in generating and refining complex reasoning chains -- a process critically dependent on effective exploration strategies. While prior work has demonstrated RLVR's empirical success, the fundamental mechanisms governing LLMs' exploration behaviors remain underexplored. This technical report presents a systematic investigation of exploration capacities "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.07534","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/2508.07534/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":"2508.07534","created_at":"2026-07-05T11:54:50.774220+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.07534v2","created_at":"2026-07-05T11:54:50.774220+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.07534","created_at":"2026-07-05T11:54:50.774220+00:00"},{"alias_kind":"pith_short_12","alias_value":"MLFKAQMAZE45","created_at":"2026-07-05T11:54:50.774220+00:00"},{"alias_kind":"pith_short_16","alias_value":"MLFKAQMAZE45MHW6","created_at":"2026-07-05T11:54:50.774220+00:00"},{"alias_kind":"pith_short_8","alias_value":"MLFKAQMA","created_at":"2026-07-05T11:54:50.774220+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.08827","citing_title":"A Survey of Reinforcement Learning for Large Reasoning Models","ref_index":106,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00380","citing_title":"ResRL: Boosting LLM Reasoning via Negative Sample Projection Residual Reinforcement Learning","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00380","citing_title":"ResRL: Boosting LLM Reasoning via Negative Sample Projection Residual Reinforcement Learning","ref_index":62,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP","json":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP.json","graph_json":"https://pith.science/api/pith-number/MLFKAQMAZE45MHW6FGHQKZ2CCP/graph.json","events_json":"https://pith.science/api/pith-number/MLFKAQMAZE45MHW6FGHQKZ2CCP/events.json","paper":"https://pith.science/paper/MLFKAQMA"},"agent_actions":{"view_html":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP","download_json":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP.json","view_paper":"https://pith.science/paper/MLFKAQMA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.07534&json=true","fetch_graph":"https://pith.science/api/pith-number/MLFKAQMAZE45MHW6FGHQKZ2CCP/graph.json","fetch_events":"https://pith.science/api/pith-number/MLFKAQMAZE45MHW6FGHQKZ2CCP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP/action/storage_attestation","attest_author":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP/action/author_attestation","sign_citation":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP/action/citation_signature","submit_replication":"https://pith.science/pith/MLFKAQMAZE45MHW6FGHQKZ2CCP/action/replication_record"}},"created_at":"2026-07-05T11:54:50.774220+00:00","updated_at":"2026-07-05T11:54:50.774220+00:00"}