{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:SMIYT6JF3RT33IL33OCLGSRXKV","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":"ec7e168f367dde92704456449ca2533fa1aa887dd58db9ffcba34e5a2d3477eb","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-13T06:32:49Z","title_canon_sha256":"b19507d8506e7f744dae6da2c1140f931113ce7594dd9cc78600704295f35401"},"schema_version":"1.0","source":{"id":"2604.11056","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.11056","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"arxiv_version","alias_value":"2604.11056v2","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.11056","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_12","alias_value":"SMIYT6JF3RT3","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_16","alias_value":"SMIYT6JF3RT33IL3","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_8","alias_value":"SMIYT6JF","created_at":"2026-05-27T01:05:54Z"}],"graph_snapshots":[{"event_id":"sha256:561631c327ed5d6bc5d5cafd19170ea74185805e8a9f7be068dbb9248faefaa3","target":"graph","created_at":"2026-05-27T01:05:54Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We adapt Conditional Mutual Information to the autoregressive RLVR setting and prove that the credit a token can carry is upper-bounded by its entropy."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The adaptation of conditional mutual information to the autoregressive RLVR setting correctly captures the credit a token can carry, and that high-entropy tokens are the primary locus of reasoning improvements."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Token credit in RLVR is upper-bounded by entropy, with reasoning gains concentrated in high-entropy tokens, motivating Entropy-Aware Policy Optimization that outperforms baselines."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"The credit a token can carry in RLVR is upper-bounded by its entropy."}],"snapshot_sha256":"760faad15b650b36e68df48d88518293dae3236d59b670f2a19f13858e515ac0"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"ee8768e998565f82a9a18e81fffef09cde0faf3c6ef275744b7e36c13208b9c3"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2604.11056/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement Learning with Verifiable Rewards (RLVR) improves the reasoning ability of Large Language Models (LLMs), but sparse outcome rewards make token-level credit assignment difficult.\n  We study token-level credit as a reward-conditioned shift from the behavior policy to a hindsight posterior.\n  In autoregressive RLVR, this shift can be expressed through Conditional Mutual Information (CMI), which shows that token entropy upper-bounds possible hindsight credit.\n  Entropy, however, indicates capacity rather than update direction, so we introduce the Four Quadrant Decomposition to separat","authors_text":"Hange Zhou, Haodong Wu, Hongyu Ge, Keyi Wu, Qihong Lin, Siyi Liu, Yongqi Zhang, Yuhang He, Zhuo Zheng, Zixin Zhong","cross_cats":["cs.AI"],"headline":"The credit a token can carry in RLVR is upper-bounded by its entropy.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-13T06:32:49Z","title":"Where Hindsight Credit Can Reside: A Signed-Capacity View of Token Updates in RLVR"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.11056","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-10T15:29:49.448633Z","id":"232c2a53-2d87-48e3-9350-a884ce6b167d","model_set":{"reader":"grok-4.3"},"one_line_summary":"Token credit in RLVR is upper-bounded by entropy, with reasoning gains concentrated in high-entropy tokens, motivating Entropy-Aware Policy Optimization that outperforms baselines.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"The credit a token can carry in RLVR is upper-bounded by its entropy.","strongest_claim":"We adapt Conditional Mutual Information to the autoregressive RLVR setting and prove that the credit a token can carry is upper-bounded by its entropy.","weakest_assumption":"The adaptation of conditional mutual information to the autoregressive RLVR setting correctly captures the credit a token can carry, and that high-entropy tokens are the primary locus of reasoning improvements."}},"verdict_id":"232c2a53-2d87-48e3-9350-a884ce6b167d"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e7ddb7aeaf9c9eaa0b30c3478fcf6c8374f8895efd8ccd5caac88000c2e790eb","target":"record","created_at":"2026-05-27T01:05:54Z","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":"ec7e168f367dde92704456449ca2533fa1aa887dd58db9ffcba34e5a2d3477eb","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-13T06:32:49Z","title_canon_sha256":"b19507d8506e7f744dae6da2c1140f931113ce7594dd9cc78600704295f35401"},"schema_version":"1.0","source":{"id":"2604.11056","kind":"arxiv","version":2}},"canonical_sha256":"931189f925dc67bda17bdb84b34a3755589b0d877512ba1d10f86db0cb49d7d3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"931189f925dc67bda17bdb84b34a3755589b0d877512ba1d10f86db0cb49d7d3","first_computed_at":"2026-05-27T01:05:54.301633Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-27T01:05:54.301633Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"vwfCiZeGKxZHUpTnz1vUeAUsENuIo6eCEsfedTCeV2tu5AEC90PKO7p9KtY4Ks0CyshQsFxBnAA64rJcX1wQCw==","signature_status":"signed_v1","signed_at":"2026-05-27T01:05:54.302481Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.11056","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e7ddb7aeaf9c9eaa0b30c3478fcf6c8374f8895efd8ccd5caac88000c2e790eb","sha256:561631c327ed5d6bc5d5cafd19170ea74185805e8a9f7be068dbb9248faefaa3"],"state_sha256":"3a77123aa5e27a5dcd1ba379d35954684ed180c3b20d6e954f68dfc5331ff7bb"}