{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5DGV25FM5YI3HFE6MGLUINXHTS","short_pith_number":"pith:5DGV25FM","schema_version":"1.0","canonical_sha256":"e8cd5d74acee11b3949e61974436e79ca6efd4823d713f022a21b1ae97bd01eb","source":{"kind":"arxiv","id":"2605.30789","version":1},"attestation_state":"computed","paper":{"title":"Smaller Models are Natural Explorers for Policy-Level Diversity in GRPO","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chufan Shi, Dingdong Wang, Junjie Wang, Ruihang Chu, Tianhe Wu, Yiming Ren, Yiran Xu, Yujiu Yang, Yukang Chen, Yu Qiao, Zicheng Lin","submitted_at":"2026-05-29T03:25:56Z","abstract_excerpt":"We identify a new dimension for enhancing rollout diversity in Group Relative Policy Optimization (GRPO) for LLMs. While GRPO relies on diverse rollouts, prevailing strategies primarily increase diversity by injecting more token-level randomness, which may introduce step-wise noise and lead to incoherent trajectories. We uncover that smaller models within the same model family inherently exhibit higher policy-level diversity, indicated by their superior pass@k relative to larger counterparts as sample counts increase. Unlike token-level noise, this diversity is temporally correlated, preserves"},"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":"2605.30789","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T03:25:56Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"77ebb098cfab7a08a4566e408a59ea6945e4319467d22915ec429b256c28a81c","abstract_canon_sha256":"2383eeed4be5b7d8b47bc6aede7d5c57292971d96b2c69131d8c87b761d70e60"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-01T01:03:16.750989Z","signature_b64":"deefPeqgXB81TWpPoewjvsNxx7YYGovWSBb/coPBzc/biwn2WGDsCTBVl2y4Q6R0Jni7fpwKwEGevEtVkfcUBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e8cd5d74acee11b3949e61974436e79ca6efd4823d713f022a21b1ae97bd01eb","last_reissued_at":"2026-06-01T01:03:16.750284Z","signature_status":"signed_v1","first_computed_at":"2026-06-01T01:03:16.750284Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Smaller Models are Natural Explorers for Policy-Level Diversity in GRPO","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Chufan Shi, Dingdong Wang, Junjie Wang, Ruihang Chu, Tianhe Wu, Yiming Ren, Yiran Xu, Yujiu Yang, Yukang Chen, Yu Qiao, Zicheng Lin","submitted_at":"2026-05-29T03:25:56Z","abstract_excerpt":"We identify a new dimension for enhancing rollout diversity in Group Relative Policy Optimization (GRPO) for LLMs. While GRPO relies on diverse rollouts, prevailing strategies primarily increase diversity by injecting more token-level randomness, which may introduce step-wise noise and lead to incoherent trajectories. We uncover that smaller models within the same model family inherently exhibit higher policy-level diversity, indicated by their superior pass@k relative to larger counterparts as sample counts increase. Unlike token-level noise, this diversity is temporally correlated, preserves"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.30789","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/2605.30789/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":"2605.30789","created_at":"2026-06-01T01:03:16.750395+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.30789v1","created_at":"2026-06-01T01:03:16.750395+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.30789","created_at":"2026-06-01T01:03:16.750395+00:00"},{"alias_kind":"pith_short_12","alias_value":"5DGV25FM5YI3","created_at":"2026-06-01T01:03:16.750395+00:00"},{"alias_kind":"pith_short_16","alias_value":"5DGV25FM5YI3HFE6","created_at":"2026-06-01T01:03:16.750395+00:00"},{"alias_kind":"pith_short_8","alias_value":"5DGV25FM","created_at":"2026-06-01T01:03:16.750395+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS","json":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS.json","graph_json":"https://pith.science/api/pith-number/5DGV25FM5YI3HFE6MGLUINXHTS/graph.json","events_json":"https://pith.science/api/pith-number/5DGV25FM5YI3HFE6MGLUINXHTS/events.json","paper":"https://pith.science/paper/5DGV25FM"},"agent_actions":{"view_html":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS","download_json":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS.json","view_paper":"https://pith.science/paper/5DGV25FM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.30789&json=true","fetch_graph":"https://pith.science/api/pith-number/5DGV25FM5YI3HFE6MGLUINXHTS/graph.json","fetch_events":"https://pith.science/api/pith-number/5DGV25FM5YI3HFE6MGLUINXHTS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS/action/storage_attestation","attest_author":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS/action/author_attestation","sign_citation":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS/action/citation_signature","submit_replication":"https://pith.science/pith/5DGV25FM5YI3HFE6MGLUINXHTS/action/replication_record"}},"created_at":"2026-06-01T01:03:16.750395+00:00","updated_at":"2026-06-01T01:03:16.750395+00:00"}