{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:GV654GS4MAYGKUVS7VECNKFZZL","short_pith_number":"pith:GV654GS4","schema_version":"1.0","canonical_sha256":"357dde1a5c60306552b2fd4826a8b9cac98f8f07a963056439244d3ddf053407","source":{"kind":"arxiv","id":"2602.01970","version":2},"attestation_state":"computed","paper":{"title":"Small Generalizable Prompt Predictive Models Can Steer Efficient RL Post-Training of Large Reasoning Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Clive Bai, Heming Zou, Kai Yang, Qi Wang, Saiyong Yang, Weijie Liu, Xiangyang Ji, Yangkun Chen, Yixiu Mao, Yuhang Jiang, Yun Qu","submitted_at":"2026-02-02T11:24:36Z","abstract_excerpt":"Reinforcement learning enhances the reasoning capabilities of large language models but often involves high computational costs due to rollout-intensive optimization. Online prompt selection presents a plausible solution by prioritizing informative prompts to improve training efficiency. However, current methods either depend on costly, exact evaluations or construct prompt-specific predictive models lacking generalization across prompts. This study introduces Generalizable Predictive Prompt Selection (GPS), which performs Bayesian inference towards prompt difficulty using a lightweight genera"},"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":"2602.01970","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-02-02T11:24:36Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"b7af0c794dc08e86973226d4b801c5aafc7499a79bc128f990975fbbdfb471c8","abstract_canon_sha256":"21adb5a9b9442b0b1c1f052da710b88ca2b4b175ea5ebc1c1aac7b1cc93a86fe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:00:33.253246Z","signature_b64":"sK63+nKBq5i293qmlcggb6dA8goibtc4FCkkx2ztECds2t0GuZwVz3sDGWS8eE0BrYaDe71BBvwZkIS3lkDeCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"357dde1a5c60306552b2fd4826a8b9cac98f8f07a963056439244d3ddf053407","last_reissued_at":"2026-05-20T00:00:33.252513Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:00:33.252513Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Small Generalizable Prompt Predictive Models Can Steer Efficient RL Post-Training of Large Reasoning Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Clive Bai, Heming Zou, Kai Yang, Qi Wang, Saiyong Yang, Weijie Liu, Xiangyang Ji, Yangkun Chen, Yixiu Mao, Yuhang Jiang, Yun Qu","submitted_at":"2026-02-02T11:24:36Z","abstract_excerpt":"Reinforcement learning enhances the reasoning capabilities of large language models but often involves high computational costs due to rollout-intensive optimization. Online prompt selection presents a plausible solution by prioritizing informative prompts to improve training efficiency. However, current methods either depend on costly, exact evaluations or construct prompt-specific predictive models lacking generalization across prompts. This study introduces Generalizable Predictive Prompt Selection (GPS), which performs Bayesian inference towards prompt difficulty using a lightweight genera"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.01970","kind":"arxiv","version":2},"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/2602.01970/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":"2602.01970","created_at":"2026-05-20T00:00:33.252635+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.01970v2","created_at":"2026-05-20T00:00:33.252635+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.01970","created_at":"2026-05-20T00:00:33.252635+00:00"},{"alias_kind":"pith_short_12","alias_value":"GV654GS4MAYG","created_at":"2026-05-20T00:00:33.252635+00:00"},{"alias_kind":"pith_short_16","alias_value":"GV654GS4MAYGKUVS","created_at":"2026-05-20T00:00:33.252635+00:00"},{"alias_kind":"pith_short_8","alias_value":"GV654GS4","created_at":"2026-05-20T00:00:33.252635+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2605.06139","citing_title":"Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06139","citing_title":"Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07114","citing_title":"Where to Spend Rollouts: Hit-Utility Optimal Rollout Allocation for Group-Based RLVR","ref_index":10,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL","json":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL.json","graph_json":"https://pith.science/api/pith-number/GV654GS4MAYGKUVS7VECNKFZZL/graph.json","events_json":"https://pith.science/api/pith-number/GV654GS4MAYGKUVS7VECNKFZZL/events.json","paper":"https://pith.science/paper/GV654GS4"},"agent_actions":{"view_html":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL","download_json":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL.json","view_paper":"https://pith.science/paper/GV654GS4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.01970&json=true","fetch_graph":"https://pith.science/api/pith-number/GV654GS4MAYGKUVS7VECNKFZZL/graph.json","fetch_events":"https://pith.science/api/pith-number/GV654GS4MAYGKUVS7VECNKFZZL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL/action/storage_attestation","attest_author":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL/action/author_attestation","sign_citation":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL/action/citation_signature","submit_replication":"https://pith.science/pith/GV654GS4MAYGKUVS7VECNKFZZL/action/replication_record"}},"created_at":"2026-05-20T00:00:33.252635+00:00","updated_at":"2026-05-20T00:00:33.252635+00:00"}