{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:YEARNEYO5GB67UV6GCCJHHABEA","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"59e93e92fa3b2002c6220795978e1fc2afe0cb455d1b493cfec60b0cfa932f3b","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-28T06:51:45Z","title_canon_sha256":"3140bf8c7d3db3ea825b157a62aaa5d31638303a99b7193f4498e9f8603775df"},"schema_version":"1.0","source":{"id":"2606.29238","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.29238","created_at":"2026-06-30T01:17:58Z"},{"alias_kind":"arxiv_version","alias_value":"2606.29238v1","created_at":"2026-06-30T01:17:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.29238","created_at":"2026-06-30T01:17:58Z"},{"alias_kind":"pith_short_12","alias_value":"YEARNEYO5GB6","created_at":"2026-06-30T01:17:58Z"},{"alias_kind":"pith_short_16","alias_value":"YEARNEYO5GB67UV6","created_at":"2026-06-30T01:17:58Z"},{"alias_kind":"pith_short_8","alias_value":"YEARNEYO","created_at":"2026-06-30T01:17:58Z"}],"graph_snapshots":[{"event_id":"sha256:e0b1e091ceaac127430a05fcfb3d1ee15f8127ff9e2daf5d2f158d12b9604611","target":"graph","created_at":"2026-06-30T01:17:58Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2606.29238/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Group Relative Policy Optimization (GRPO) eliminates the learned critic in PPO by using the mean reward of grouped rollouts as a baseline. We provide a rigorous derivation of GRPO from first principles of the policy gradient theorem, revealing a fundamental credit assignment failure: under output-only reward, every token in a rollout receives identical advantage, collapsing token-level credit to a single scalar. We prove this induces gradient sparsity that intensifies over training, and demonstrate empirically via SVD analysis of GRPO gradients on Nemotron-4B/GSM8K that the gradient matrix has","authors_text":"Amritansh Mishra, Berkcan Kapusuzoglu, Supriyo Chakraborty","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-28T06:51:45Z","title":"On the Policy Gradient Foundations of Group Relative Policy Optimization: Credit Assignment, Gradient Sparsity, and Rank Collapse"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.29238","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:97fbea4471e6f9aa6ffacb80d86b274647ff88780919fd819b4f2f24b79aa501","target":"record","created_at":"2026-06-30T01:17:58Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"59e93e92fa3b2002c6220795978e1fc2afe0cb455d1b493cfec60b0cfa932f3b","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-06-28T06:51:45Z","title_canon_sha256":"3140bf8c7d3db3ea825b157a62aaa5d31638303a99b7193f4498e9f8603775df"},"schema_version":"1.0","source":{"id":"2606.29238","kind":"arxiv","version":1}},"canonical_sha256":"c10116930ee983efd2be3084939c01202fd5e7de134b1b7a1179abfdc94121ff","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c10116930ee983efd2be3084939c01202fd5e7de134b1b7a1179abfdc94121ff","first_computed_at":"2026-06-30T01:17:58.525942Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-30T01:17:58.525942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"8NyNCqIYjh+3zd00ikLfXZkndFVJmMso7Fj8mgwkyxJlw8zS5NNSbOVpRAUit6wMmeowahZk/deoP0JmR7oPCg==","signature_status":"signed_v1","signed_at":"2026-06-30T01:17:58.526637Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.29238","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:97fbea4471e6f9aa6ffacb80d86b274647ff88780919fd819b4f2f24b79aa501","sha256:e0b1e091ceaac127430a05fcfb3d1ee15f8127ff9e2daf5d2f158d12b9604611"],"state_sha256":"dfcba751c9d18ef72a55ec30fd52e5541141ea5c374c45de4f57baab0ba9eae9"}