{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XKB76IJKJS464ULME7OLT2RTM2","short_pith_number":"pith:XKB76IJK","schema_version":"1.0","canonical_sha256":"ba83ff212a4cb9ee516c27dcb9ea3366837e58ae8f70076a7f4c7e7c270cfec4","source":{"kind":"arxiv","id":"2601.07408","version":2},"attestation_state":"computed","paper":{"title":"Outcome-Grounded Advantage Reshaping for Fine-Grained Credit Assignment in Mathematical Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Deqing Yang, Feng Xiao, Hongcheng Guo, Liu Kang, Luxi Xing, Qingyi Si, Weikang Gong, Yanghua Xiao, Zhuoran Li, Ziheng Li","submitted_at":"2026-01-12T10:48:02Z","abstract_excerpt":"Group Relative Policy Optimization (GRPO) has emerged as a promising critic-free reinforcement learning paradigm for reasoning tasks. However, standard GRPO employs a coarse-grained credit assignment mechanism that propagates group-level rewards uniformly to to every token in a sequence, neglecting the varying contribution of individual reasoning steps. We address this limitation by introducing Outcome-grounded Advantage Reshaping (OAR), a fine-grained credit assignment mechanism that redistributes advantages based on how much each token influences the model's final answer. We instantiate OAR "},"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":"2601.07408","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-01-12T10:48:02Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e1b4eb71ca227ad51de8f75cc8e6f63a2c73dc9892cca58185aba5f4aec51732","abstract_canon_sha256":"c21ee5f1d411588d3d65e2ee4249eacb17d11ec450ef8b92f9212be5ad3383de"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-04T01:08:39.737639Z","signature_b64":"JE++Q0hupVpNhg7rqJnLWMSDpznWLtS1xtCAvvJMe+52/9msK7HuC5uBV/Ta0LVmBBc7X0k2dS04+Hq0odkeAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ba83ff212a4cb9ee516c27dcb9ea3366837e58ae8f70076a7f4c7e7c270cfec4","last_reissued_at":"2026-06-04T01:08:39.736778Z","signature_status":"signed_v1","first_computed_at":"2026-06-04T01:08:39.736778Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Outcome-Grounded Advantage Reshaping for Fine-Grained Credit Assignment in Mathematical Reasoning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Deqing Yang, Feng Xiao, Hongcheng Guo, Liu Kang, Luxi Xing, Qingyi Si, Weikang Gong, Yanghua Xiao, Zhuoran Li, Ziheng Li","submitted_at":"2026-01-12T10:48:02Z","abstract_excerpt":"Group Relative Policy Optimization (GRPO) has emerged as a promising critic-free reinforcement learning paradigm for reasoning tasks. However, standard GRPO employs a coarse-grained credit assignment mechanism that propagates group-level rewards uniformly to to every token in a sequence, neglecting the varying contribution of individual reasoning steps. We address this limitation by introducing Outcome-grounded Advantage Reshaping (OAR), a fine-grained credit assignment mechanism that redistributes advantages based on how much each token influences the model's final answer. We instantiate OAR "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.07408","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/2601.07408/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":"2601.07408","created_at":"2026-06-04T01:08:39.736902+00:00"},{"alias_kind":"arxiv_version","alias_value":"2601.07408v2","created_at":"2026-06-04T01:08:39.736902+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.07408","created_at":"2026-06-04T01:08:39.736902+00:00"},{"alias_kind":"pith_short_12","alias_value":"XKB76IJKJS46","created_at":"2026-06-04T01:08:39.736902+00:00"},{"alias_kind":"pith_short_16","alias_value":"XKB76IJKJS464ULM","created_at":"2026-06-04T01:08:39.736902+00:00"},{"alias_kind":"pith_short_8","alias_value":"XKB76IJK","created_at":"2026-06-04T01:08:39.736902+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2605.19447","citing_title":"What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2604.03128","citing_title":"Self-Distilled RLVR","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2604.23318","citing_title":"Hidden States Know Where Reasoning Diverges: Credit Assignment via Span-Level Wasserstein Distance","ref_index":15,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2","json":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2.json","graph_json":"https://pith.science/api/pith-number/XKB76IJKJS464ULME7OLT2RTM2/graph.json","events_json":"https://pith.science/api/pith-number/XKB76IJKJS464ULME7OLT2RTM2/events.json","paper":"https://pith.science/paper/XKB76IJK"},"agent_actions":{"view_html":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2","download_json":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2.json","view_paper":"https://pith.science/paper/XKB76IJK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2601.07408&json=true","fetch_graph":"https://pith.science/api/pith-number/XKB76IJKJS464ULME7OLT2RTM2/graph.json","fetch_events":"https://pith.science/api/pith-number/XKB76IJKJS464ULME7OLT2RTM2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2/action/storage_attestation","attest_author":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2/action/author_attestation","sign_citation":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2/action/citation_signature","submit_replication":"https://pith.science/pith/XKB76IJKJS464ULME7OLT2RTM2/action/replication_record"}},"created_at":"2026-06-04T01:08:39.736902+00:00","updated_at":"2026-06-04T01:08:39.736902+00:00"}