{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KZWMQ6ZCSHPAJFEZENQT3YJSDU","short_pith_number":"pith:KZWMQ6ZC","schema_version":"1.0","canonical_sha256":"566cc87b2291de04949923613de1321d349f1615077dad308bb42856cdc22a33","source":{"kind":"arxiv","id":"2507.02841","version":1},"attestation_state":"computed","paper":{"title":"StepHint: Multi-level Stepwise Hints Enhance Reinforcement Learning to Reason","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Ang Lv, Feng Wang, Haoyuan Hu, Jinpeng Li, Kaiyi Zhang, Rui Yan, Yongbo Wang","submitted_at":"2025-07-03T17:51:06Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) is a promising approach for improving the complex reasoning abilities of large language models (LLMs). However, current RLVR methods face two significant challenges: the near-miss reward problem, where a small mistake can invalidate an otherwise correct reasoning process, greatly hindering training efficiency; and exploration stagnation, where models tend to focus on solutions within their ``comfort zone,'' lacking the motivation to explore potentially more effective alternatives. To address these challenges, we propose StepHint, a novel RL"},"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":"2507.02841","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.AI","submitted_at":"2025-07-03T17:51:06Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"b990ed007b06d871a93856b91699ac10e60786532f4781394de2130d67a631fe","abstract_canon_sha256":"7071725cc6d821357e5e666e6d523658e9ff6ce3d2bd390cdb81fe11b6fed664"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:31:35.237416Z","signature_b64":"sHvgArx7E/t9LDDKZ+3wEALSeizLZS4Kjk8uq56SkbH31255zAGHPRyrcihhCEBiOLwNL/Mom5HTHRnginBoCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"566cc87b2291de04949923613de1321d349f1615077dad308bb42856cdc22a33","last_reissued_at":"2026-07-05T11:31:35.236895Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:31:35.236895Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"StepHint: Multi-level Stepwise Hints Enhance Reinforcement Learning to Reason","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Ang Lv, Feng Wang, Haoyuan Hu, Jinpeng Li, Kaiyi Zhang, Rui Yan, Yongbo Wang","submitted_at":"2025-07-03T17:51:06Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) is a promising approach for improving the complex reasoning abilities of large language models (LLMs). However, current RLVR methods face two significant challenges: the near-miss reward problem, where a small mistake can invalidate an otherwise correct reasoning process, greatly hindering training efficiency; and exploration stagnation, where models tend to focus on solutions within their ``comfort zone,'' lacking the motivation to explore potentially more effective alternatives. To address these challenges, we propose StepHint, a novel RL"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.02841","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/2507.02841/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":"2507.02841","created_at":"2026-07-05T11:31:35.236956+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.02841v1","created_at":"2026-07-05T11:31:35.236956+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.02841","created_at":"2026-07-05T11:31:35.236956+00:00"},{"alias_kind":"pith_short_12","alias_value":"KZWMQ6ZCSHPA","created_at":"2026-07-05T11:31:35.236956+00:00"},{"alias_kind":"pith_short_16","alias_value":"KZWMQ6ZCSHPAJFEZ","created_at":"2026-07-05T11:31:35.236956+00:00"},{"alias_kind":"pith_short_8","alias_value":"KZWMQ6ZC","created_at":"2026-07-05T11:31:35.236956+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25450","citing_title":"The Generalization Spectrum: A Chromatographic Approach to Evaluating Learning Algorithms","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25450","citing_title":"The Generalization Spectrum: A Chromatographic Approach to Evaluating Learning Algorithms","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18216","citing_title":"Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13680","citing_title":"Learning to Reason by Analogy via Retrieval-Augmented Reinforcement Fine-Tuning","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22567","citing_title":"LANG: Reinforcement Learning for Multilingual Reasoning with Language-Adaptive Hint Guidance","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21467","citing_title":"DelTA: Discriminative Token Credit Assignment for Reinforcement Learning from Verifiable Rewards","ref_index":80,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05112","citing_title":"Rollout Pass-Rate Control: Steering Binary-Reward RL Toward Its Most Informative Regime","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11636","citing_title":"Seir\\^enes: Adversarial Self-Play with Evolving Distractions for LLM Reasoning","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05112","citing_title":"Rollout Pass-Rate Control: Steering Binary-Reward RL Toward Its Most Informative Regime","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05112","citing_title":"Rollout Pass-Rate Control: Steering Binary-Reward RL Toward Its Most Informative Regime","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU","json":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU.json","graph_json":"https://pith.science/api/pith-number/KZWMQ6ZCSHPAJFEZENQT3YJSDU/graph.json","events_json":"https://pith.science/api/pith-number/KZWMQ6ZCSHPAJFEZENQT3YJSDU/events.json","paper":"https://pith.science/paper/KZWMQ6ZC"},"agent_actions":{"view_html":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU","download_json":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU.json","view_paper":"https://pith.science/paper/KZWMQ6ZC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.02841&json=true","fetch_graph":"https://pith.science/api/pith-number/KZWMQ6ZCSHPAJFEZENQT3YJSDU/graph.json","fetch_events":"https://pith.science/api/pith-number/KZWMQ6ZCSHPAJFEZENQT3YJSDU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU/action/storage_attestation","attest_author":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU/action/author_attestation","sign_citation":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU/action/citation_signature","submit_replication":"https://pith.science/pith/KZWMQ6ZCSHPAJFEZENQT3YJSDU/action/replication_record"}},"created_at":"2026-07-05T11:31:35.236956+00:00","updated_at":"2026-07-05T11:31:35.236956+00:00"}