{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:IVEMD2RMDPFM7CZY6Z2MNVVYDL","short_pith_number":"pith:IVEMD2RM","canonical_record":{"source":{"id":"2509.21154","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-25T13:40:36Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8ccdd1feaf668166a1a097e72641217affdce7de6a4405524820638bb5abce00","abstract_canon_sha256":"568cb2708829e3bd83c602f2ca9d8ffb72a97e94dbe41b88593c72e057b730cb"},"schema_version":"1.0"},"canonical_sha256":"4548c1ea2c1bcacf8b38f674c6d6b81ae7c1f5c95de198135a1baf56c09bf5a9","source":{"kind":"arxiv","id":"2509.21154","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.21154","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"arxiv_version","alias_value":"2509.21154v4","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.21154","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"pith_short_12","alias_value":"IVEMD2RMDPFM","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"pith_short_16","alias_value":"IVEMD2RMDPFM7CZY","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"pith_short_8","alias_value":"IVEMD2RM","created_at":"2026-05-29T02:05:36Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:IVEMD2RMDPFM7CZY6Z2MNVVYDL","target":"record","payload":{"canonical_record":{"source":{"id":"2509.21154","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-25T13:40:36Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8ccdd1feaf668166a1a097e72641217affdce7de6a4405524820638bb5abce00","abstract_canon_sha256":"568cb2708829e3bd83c602f2ca9d8ffb72a97e94dbe41b88593c72e057b730cb"},"schema_version":"1.0"},"canonical_sha256":"4548c1ea2c1bcacf8b38f674c6d6b81ae7c1f5c95de198135a1baf56c09bf5a9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:05:36.366015Z","signature_b64":"YnnTrNTEW8IKS4i+5Z6E5w9JTyLo1GU4zUqRtZGcPCxqthrc5Mjvh26aLgLGnr2uY6v1DO3Cx1KgskX9y5oqAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4548c1ea2c1bcacf8b38f674c6d6b81ae7c1f5c95de198135a1baf56c09bf5a9","last_reissued_at":"2026-05-29T02:05:36.365588Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:05:36.365588Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2509.21154","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-29T02:05:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8yEnulgC9F0pplq2Dcrogdlyk0Z6R2D9M67aQBKHwL2zguo7xz8fM1gFWQqhp3+rdFMaKtTYVgqtGXX5/ZaOCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T07:51:54.483851Z"},"content_sha256":"e161b9e8e4746cc606baf73bd7260f9427e541a54820acd04b8a8933a226cb07","schema_version":"1.0","event_id":"sha256:e161b9e8e4746cc606baf73bd7260f9427e541a54820acd04b8a8933a226cb07"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:IVEMD2RMDPFM7CZY6Z2MNVVYDL","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"GRPO is Secretly a Process Reward Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Alexander Koller, Michael Sullivan","submitted_at":"2025-09-25T13:40:36Z","abstract_excerpt":"Process reward models (PRMs) allow for fine-grained credit assignment in reinforcement learning (RL), and seemingly contrast with outcome reward models (ORMs), which assign a single reward to an entire trajectory. However, we provide theoretical proof in this work that the Group Relative Policy Optimization (GRPO) RL algorithm equipped with an ORM is in fact equivalent to a PRM-aware RL objective equipped with a non-trivial, Monte-Carlo-based PRM (given mild assumptions). Leveraging the framework of GRPO-as-a-PRM, we identify a flaw in the GRPO objective that interacts with imbalanced process "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.21154","kind":"arxiv","version":4},"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/2509.21154/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-29T02:05:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xE0nFkugVNDhZwKwlDmhNkeyemVJYHAjKwsC43xsz95waE0837GNgmB1pxCBpzoa1LsF8V/0Hv/RwhftbY5TCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T07:51:54.484409Z"},"content_sha256":"63655eb5c8d160e4c687a3c3e0f57cc8edb475c1491d9ed85d087ebb4e2664ca","schema_version":"1.0","event_id":"sha256:63655eb5c8d160e4c687a3c3e0f57cc8edb475c1491d9ed85d087ebb4e2664ca"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/IVEMD2RMDPFM7CZY6Z2MNVVYDL/bundle.json","state_url":"https://pith.science/pith/IVEMD2RMDPFM7CZY6Z2MNVVYDL/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/IVEMD2RMDPFM7CZY6Z2MNVVYDL/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-02T07:51:54Z","links":{"resolver":"https://pith.science/pith/IVEMD2RMDPFM7CZY6Z2MNVVYDL","bundle":"https://pith.science/pith/IVEMD2RMDPFM7CZY6Z2MNVVYDL/bundle.json","state":"https://pith.science/pith/IVEMD2RMDPFM7CZY6Z2MNVVYDL/state.json","well_known_bundle":"https://pith.science/.well-known/pith/IVEMD2RMDPFM7CZY6Z2MNVVYDL/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:IVEMD2RMDPFM7CZY6Z2MNVVYDL","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":"568cb2708829e3bd83c602f2ca9d8ffb72a97e94dbe41b88593c72e057b730cb","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-25T13:40:36Z","title_canon_sha256":"8ccdd1feaf668166a1a097e72641217affdce7de6a4405524820638bb5abce00"},"schema_version":"1.0","source":{"id":"2509.21154","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.21154","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"arxiv_version","alias_value":"2509.21154v4","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.21154","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"pith_short_12","alias_value":"IVEMD2RMDPFM","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"pith_short_16","alias_value":"IVEMD2RMDPFM7CZY","created_at":"2026-05-29T02:05:36Z"},{"alias_kind":"pith_short_8","alias_value":"IVEMD2RM","created_at":"2026-05-29T02:05:36Z"}],"graph_snapshots":[{"event_id":"sha256:63655eb5c8d160e4c687a3c3e0f57cc8edb475c1491d9ed85d087ebb4e2664ca","target":"graph","created_at":"2026-05-29T02:05:36Z","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/2509.21154/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Process reward models (PRMs) allow for fine-grained credit assignment in reinforcement learning (RL), and seemingly contrast with outcome reward models (ORMs), which assign a single reward to an entire trajectory. However, we provide theoretical proof in this work that the Group Relative Policy Optimization (GRPO) RL algorithm equipped with an ORM is in fact equivalent to a PRM-aware RL objective equipped with a non-trivial, Monte-Carlo-based PRM (given mild assumptions). Leveraging the framework of GRPO-as-a-PRM, we identify a flaw in the GRPO objective that interacts with imbalanced process ","authors_text":"Alexander Koller, Michael Sullivan","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-25T13:40:36Z","title":"GRPO is Secretly a Process Reward Model"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.21154","kind":"arxiv","version":4},"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:e161b9e8e4746cc606baf73bd7260f9427e541a54820acd04b8a8933a226cb07","target":"record","created_at":"2026-05-29T02:05:36Z","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":"568cb2708829e3bd83c602f2ca9d8ffb72a97e94dbe41b88593c72e057b730cb","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-25T13:40:36Z","title_canon_sha256":"8ccdd1feaf668166a1a097e72641217affdce7de6a4405524820638bb5abce00"},"schema_version":"1.0","source":{"id":"2509.21154","kind":"arxiv","version":4}},"canonical_sha256":"4548c1ea2c1bcacf8b38f674c6d6b81ae7c1f5c95de198135a1baf56c09bf5a9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4548c1ea2c1bcacf8b38f674c6d6b81ae7c1f5c95de198135a1baf56c09bf5a9","first_computed_at":"2026-05-29T02:05:36.365588Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-29T02:05:36.365588Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"YnnTrNTEW8IKS4i+5Z6E5w9JTyLo1GU4zUqRtZGcPCxqthrc5Mjvh26aLgLGnr2uY6v1DO3Cx1KgskX9y5oqAw==","signature_status":"signed_v1","signed_at":"2026-05-29T02:05:36.366015Z","signed_message":"canonical_sha256_bytes"},"source_id":"2509.21154","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e161b9e8e4746cc606baf73bd7260f9427e541a54820acd04b8a8933a226cb07","sha256:63655eb5c8d160e4c687a3c3e0f57cc8edb475c1491d9ed85d087ebb4e2664ca"],"state_sha256":"ab4bbd30373900a955f75d00127415e5d74a8d523addc5799782901eac64bd1e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9iXzUaYxTyU9zkkGFd3BIbn9jojyOs3g3cr8vg8skNEE8wvHtz2+KJNrwUZNQ5I4zApTCl357wAKx24pIRaYAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-02T07:51:54.486404Z","bundle_sha256":"e285950222389ed14383387e13f39f0f8c5f0a7dafda3429cbe4d7f67338b52a"}}