{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:EUBTURJBWCDKBJ4Z4TXGUOJCDO","short_pith_number":"pith:EUBTURJB","schema_version":"1.0","canonical_sha256":"25033a4521b086a0a799e4ee6a39221b8fd47b217ad6538afdc5f97ee11fea84","source":{"kind":"arxiv","id":"2505.02686","version":2},"attestation_state":"computed","paper":{"title":"Sailing by the Stars: A Survey on Reward Models and Learning Strategies for Learning from Rewards","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Xiaobao Wu","submitted_at":"2025-05-05T14:33:49Z","abstract_excerpt":"Recent developments in Large Language Models (LLMs) have shifted from pre-training scaling to post-training and test-time scaling. Across these developments, a key unified paradigm has arisen: Learning from Rewards, where reward signals act as the guiding stars to steer LLM behavior. It has underpinned a wide range of prevalent techniques, such as reinforcement learning (RLHF, RLAIF, DPO, and GRPO), reward-guided decoding, and post-hoc correction. Crucially, this paradigm enables the transition from passive learning from static data to active learning from dynamic feedback. This endows LLMs wi"},"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":"2505.02686","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-05T14:33:49Z","cross_cats_sorted":[],"title_canon_sha256":"6162642b0c937f7f635d360a7d0dc0a4bb76e22a4c35dd038a4a67f619943b11","abstract_canon_sha256":"15c4e6ed20ef6f0cf2efa89aad5c764fd1fe4912623bdb45825794bcc2bb24bc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:20:38.266410Z","signature_b64":"EuUGD0x/CYASqJv+mh6AfV7UIhF4DSvdQjsSHD/BK/jZfws29+SP4K21Ns5X1fyWukJQwHLxZY6nxC36h5FJBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"25033a4521b086a0a799e4ee6a39221b8fd47b217ad6538afdc5f97ee11fea84","last_reissued_at":"2026-07-05T11:20:38.265912Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:20:38.265912Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sailing by the Stars: A Survey on Reward Models and Learning Strategies for Learning from Rewards","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Xiaobao Wu","submitted_at":"2025-05-05T14:33:49Z","abstract_excerpt":"Recent developments in Large Language Models (LLMs) have shifted from pre-training scaling to post-training and test-time scaling. Across these developments, a key unified paradigm has arisen: Learning from Rewards, where reward signals act as the guiding stars to steer LLM behavior. It has underpinned a wide range of prevalent techniques, such as reinforcement learning (RLHF, RLAIF, DPO, and GRPO), reward-guided decoding, and post-hoc correction. Crucially, this paradigm enables the transition from passive learning from static data to active learning from dynamic feedback. This endows LLMs wi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.02686","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/2505.02686/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":"2505.02686","created_at":"2026-07-05T11:20:38.265971+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.02686v2","created_at":"2026-07-05T11:20:38.265971+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.02686","created_at":"2026-07-05T11:20:38.265971+00:00"},{"alias_kind":"pith_short_12","alias_value":"EUBTURJBWCDK","created_at":"2026-07-05T11:20:38.265971+00:00"},{"alias_kind":"pith_short_16","alias_value":"EUBTURJBWCDKBJ4Z","created_at":"2026-07-05T11:20:38.265971+00:00"},{"alias_kind":"pith_short_8","alias_value":"EUBTURJB","created_at":"2026-07-05T11:20:38.265971+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"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":228,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01249","citing_title":"Trust Region On-Policy Distillation","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2508.09521","citing_title":"PEER: Unified Process-Outcome Reinforcement Learning for Structured Empathetic Reasoning","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO","json":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO.json","graph_json":"https://pith.science/api/pith-number/EUBTURJBWCDKBJ4Z4TXGUOJCDO/graph.json","events_json":"https://pith.science/api/pith-number/EUBTURJBWCDKBJ4Z4TXGUOJCDO/events.json","paper":"https://pith.science/paper/EUBTURJB"},"agent_actions":{"view_html":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO","download_json":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO.json","view_paper":"https://pith.science/paper/EUBTURJB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.02686&json=true","fetch_graph":"https://pith.science/api/pith-number/EUBTURJBWCDKBJ4Z4TXGUOJCDO/graph.json","fetch_events":"https://pith.science/api/pith-number/EUBTURJBWCDKBJ4Z4TXGUOJCDO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO/action/storage_attestation","attest_author":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO/action/author_attestation","sign_citation":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO/action/citation_signature","submit_replication":"https://pith.science/pith/EUBTURJBWCDKBJ4Z4TXGUOJCDO/action/replication_record"}},"created_at":"2026-07-05T11:20:38.265971+00:00","updated_at":"2026-07-05T11:20:38.265971+00:00"}