{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:TRBKOHBTWCOH66D5OVN5PMQ3PT","short_pith_number":"pith:TRBKOHBT","canonical_record":{"source":{"id":"2509.06941","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-08T17:52:56Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"3c07b3cf4acdb81a4de9701cfc34cd7987932f0b1032721b69fcd4bc911cda12","abstract_canon_sha256":"aa80fe971f6acf719de2d54ad38016c306305cfd7038ccb8b45dd4b50e9fe356"},"schema_version":"1.0"},"canonical_sha256":"9c42a71c33b09c7f787d755bd7b21b7cc34411db260b3144076543d2d0f9c72a","source":{"kind":"arxiv","id":"2509.06941","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.06941","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"arxiv_version","alias_value":"2509.06941v1","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.06941","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"pith_short_12","alias_value":"TRBKOHBTWCOH","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"pith_short_16","alias_value":"TRBKOHBTWCOH66D5","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"pith_short_8","alias_value":"TRBKOHBT","created_at":"2026-07-05T12:06:55Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:TRBKOHBTWCOH66D5OVN5PMQ3PT","target":"record","payload":{"canonical_record":{"source":{"id":"2509.06941","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-08T17:52:56Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"3c07b3cf4acdb81a4de9701cfc34cd7987932f0b1032721b69fcd4bc911cda12","abstract_canon_sha256":"aa80fe971f6acf719de2d54ad38016c306305cfd7038ccb8b45dd4b50e9fe356"},"schema_version":"1.0"},"canonical_sha256":"9c42a71c33b09c7f787d755bd7b21b7cc34411db260b3144076543d2d0f9c72a","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:06:55.489682Z","signature_b64":"ASIK/UGEYz6Us89mARZMMnNBoH1QN7ZyKcKxaa4GGZpcFV1li1xCQGn2geMpatXDuJkqOmIgQGsaizk/GcFdDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9c42a71c33b09c7f787d755bd7b21b7cc34411db260b3144076543d2d0f9c72a","last_reissued_at":"2026-07-05T12:06:55.489232Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:06:55.489232Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2509.06941","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T12:06:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"eSAEHA3PmZdJWmH/QhGWBo3LGOzUBldMlSzIs5LnMtiuDLcTv/BzhMygY+PaxmogrXzv/qlI4gl6EejvKGR+AA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-21T20:52:44.255175Z"},"content_sha256":"3e2adebd02cfce1f1116049a94935a39796bb1c0659e3774870f75d893a2f275","schema_version":"1.0","event_id":"sha256:3e2adebd02cfce1f1116049a94935a39796bb1c0659e3774870f75d893a2f275"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:TRBKOHBTWCOH66D5OVN5PMQ3PT","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Outcome-based Exploration for LLM Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Julia Kempe, Remi Munos, Yuda Song","submitted_at":"2025-09-08T17:52:56Z","abstract_excerpt":"Reinforcement learning (RL) has emerged as a powerful method for improving the reasoning abilities of large language models (LLMs). Outcome-based RL, which rewards policies solely for the correctness of the final answer, yields substantial accuracy gains but also induces a systematic loss in generation diversity. This collapse undermines real-world performance, where diversity is critical for test-time scaling. We analyze this phenomenon by viewing RL post-training as a sampling process and show that, strikingly, RL can reduce effective diversity even on the training set relative to the base m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.06941","kind":"arxiv","version":1},"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/2509.06941/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":"2026-07-05T12:06:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"OAlFceoNLndtkWLu1o1eJoZH1f2WF2eRYZFTy5G6AZNoe9/xE6xnfOWH+IuBC4LYv60DfuNfKSTBlZbsYQW0CA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-21T20:52:44.255673Z"},"content_sha256":"c3e8f2555d23062031e4a8c2b2e3bca12a975c6542c7448e5f793611fa80d76e","schema_version":"1.0","event_id":"sha256:c3e8f2555d23062031e4a8c2b2e3bca12a975c6542c7448e5f793611fa80d76e"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TRBKOHBTWCOH66D5OVN5PMQ3PT/bundle.json","state_url":"https://pith.science/pith/TRBKOHBTWCOH66D5OVN5PMQ3PT/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TRBKOHBTWCOH66D5OVN5PMQ3PT/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":"2026-08-21T20:52:44Z","links":{"resolver":"https://pith.science/pith/TRBKOHBTWCOH66D5OVN5PMQ3PT","bundle":"https://pith.science/pith/TRBKOHBTWCOH66D5OVN5PMQ3PT/bundle.json","state":"https://pith.science/pith/TRBKOHBTWCOH66D5OVN5PMQ3PT/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TRBKOHBTWCOH66D5OVN5PMQ3PT/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:TRBKOHBTWCOH66D5OVN5PMQ3PT","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":"aa80fe971f6acf719de2d54ad38016c306305cfd7038ccb8b45dd4b50e9fe356","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-08T17:52:56Z","title_canon_sha256":"3c07b3cf4acdb81a4de9701cfc34cd7987932f0b1032721b69fcd4bc911cda12"},"schema_version":"1.0","source":{"id":"2509.06941","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2509.06941","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"arxiv_version","alias_value":"2509.06941v1","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.06941","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"pith_short_12","alias_value":"TRBKOHBTWCOH","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"pith_short_16","alias_value":"TRBKOHBTWCOH66D5","created_at":"2026-07-05T12:06:55Z"},{"alias_kind":"pith_short_8","alias_value":"TRBKOHBT","created_at":"2026-07-05T12:06:55Z"}],"graph_snapshots":[{"event_id":"sha256:c3e8f2555d23062031e4a8c2b2e3bca12a975c6542c7448e5f793611fa80d76e","target":"graph","created_at":"2026-07-05T12:06:55Z","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/2509.06941/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement learning (RL) has emerged as a powerful method for improving the reasoning abilities of large language models (LLMs). Outcome-based RL, which rewards policies solely for the correctness of the final answer, yields substantial accuracy gains but also induces a systematic loss in generation diversity. This collapse undermines real-world performance, where diversity is critical for test-time scaling. We analyze this phenomenon by viewing RL post-training as a sampling process and show that, strikingly, RL can reduce effective diversity even on the training set relative to the base m","authors_text":"Julia Kempe, Remi Munos, Yuda Song","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-08T17:52:56Z","title":"Outcome-based Exploration for LLM Reasoning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.06941","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":"sha256:3e2adebd02cfce1f1116049a94935a39796bb1c0659e3774870f75d893a2f275","target":"record","created_at":"2026-07-05T12:06:55Z","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":"aa80fe971f6acf719de2d54ad38016c306305cfd7038ccb8b45dd4b50e9fe356","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-08T17:52:56Z","title_canon_sha256":"3c07b3cf4acdb81a4de9701cfc34cd7987932f0b1032721b69fcd4bc911cda12"},"schema_version":"1.0","source":{"id":"2509.06941","kind":"arxiv","version":1}},"canonical_sha256":"9c42a71c33b09c7f787d755bd7b21b7cc34411db260b3144076543d2d0f9c72a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9c42a71c33b09c7f787d755bd7b21b7cc34411db260b3144076543d2d0f9c72a","first_computed_at":"2026-07-05T12:06:55.489232Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T12:06:55.489232Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ASIK/UGEYz6Us89mARZMMnNBoH1QN7ZyKcKxaa4GGZpcFV1li1xCQGn2geMpatXDuJkqOmIgQGsaizk/GcFdDg==","signature_status":"signed_v1","signed_at":"2026-07-05T12:06:55.489682Z","signed_message":"canonical_sha256_bytes"},"source_id":"2509.06941","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3e2adebd02cfce1f1116049a94935a39796bb1c0659e3774870f75d893a2f275","sha256:c3e8f2555d23062031e4a8c2b2e3bca12a975c6542c7448e5f793611fa80d76e"],"state_sha256":"e94bcf483f18373aeb9cd7172b2bd2ed682044750e0aa812676c0c1ea4d0b67f"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GPXMrZole4bbJsLx/Z/VHDM6Bb1+wCI1NWzcB9oZLXL/nUf/aJ8Hd8XcS93rnKHy0pJgcxX9rfZS9rwJGbVvAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-21T20:52:44.260553Z","bundle_sha256":"27cf97ccdbad4dc46d6f0dfcd3bcc955897222e1316a69204718afa171f966c7"}}