{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SDT6RVDXUNVLK7LVGN2HWPIPX7","short_pith_number":"pith:SDT6RVDX","schema_version":"1.0","canonical_sha256":"90e7e8d477a36ab57d7533747b3d0fbfc3e357ea03f7820040b221a0a819d421","source":{"kind":"arxiv","id":"2605.28109","version":1},"attestation_state":"computed","paper":{"title":"Long Live The Balance: Information Bottleneck Driven Tree-based Policy Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bin Yang, Bowen Xu, Hao Jiang, Hongtao Duan, Lulu Hu, Minying Zhang, Qihua Chen, Shurui Li, Tianpeng Bu, Xin Liu","submitted_at":"2026-05-27T08:01:42Z","abstract_excerpt":"Recent advances in online reinforcement learning (RL) for large language models (LLMs) have demonstrated promising performance in complex reasoning tasks. However, they often exhibit an imbalanced exploration-exploitation trade-off, resulting in unstable optimization and sub-optimal performance. We introduce IB-Score, a novel metric grounded in Information Bottleneck theory that evaluates policy's exploration-exploitation balance by quantifying the trade-off between step-level reasoning diversity and mutual information shared with the correct answer. Analysis based on IB-Score shows that popul"},"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":"2605.28109","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T08:01:42Z","cross_cats_sorted":[],"title_canon_sha256":"1e8f40ece78900b71e7dba7d7a65a99033b24a251acc2b031195ecd3a4bb9146","abstract_canon_sha256":"6a176e33a0ee753d947d701f5d15dca482036f1a9cce07618fcbf9d6ee581739"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T01:04:59.122210Z","signature_b64":"Ff4VcVSlfRnkfo5MsB7q0aobyRu+MkYEk9wOXI/MJ+zj5yJc8U+akc7kEEjMgitQGzMwIjEWUuWj54ido4p5AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"90e7e8d477a36ab57d7533747b3d0fbfc3e357ea03f7820040b221a0a819d421","last_reissued_at":"2026-05-28T01:04:59.121748Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T01:04:59.121748Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Long Live The Balance: Information Bottleneck Driven Tree-based Policy Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bin Yang, Bowen Xu, Hao Jiang, Hongtao Duan, Lulu Hu, Minying Zhang, Qihua Chen, Shurui Li, Tianpeng Bu, Xin Liu","submitted_at":"2026-05-27T08:01:42Z","abstract_excerpt":"Recent advances in online reinforcement learning (RL) for large language models (LLMs) have demonstrated promising performance in complex reasoning tasks. However, they often exhibit an imbalanced exploration-exploitation trade-off, resulting in unstable optimization and sub-optimal performance. We introduce IB-Score, a novel metric grounded in Information Bottleneck theory that evaluates policy's exploration-exploitation balance by quantifying the trade-off between step-level reasoning diversity and mutual information shared with the correct answer. Analysis based on IB-Score shows that popul"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.28109","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/2605.28109/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":"2605.28109","created_at":"2026-05-28T01:04:59.121813+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.28109v1","created_at":"2026-05-28T01:04:59.121813+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.28109","created_at":"2026-05-28T01:04:59.121813+00:00"},{"alias_kind":"pith_short_12","alias_value":"SDT6RVDXUNVL","created_at":"2026-05-28T01:04:59.121813+00:00"},{"alias_kind":"pith_short_16","alias_value":"SDT6RVDXUNVLK7LV","created_at":"2026-05-28T01:04:59.121813+00:00"},{"alias_kind":"pith_short_8","alias_value":"SDT6RVDX","created_at":"2026-05-28T01:04:59.121813+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/SDT6RVDXUNVLK7LVGN2HWPIPX7","json":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7.json","graph_json":"https://pith.science/api/pith-number/SDT6RVDXUNVLK7LVGN2HWPIPX7/graph.json","events_json":"https://pith.science/api/pith-number/SDT6RVDXUNVLK7LVGN2HWPIPX7/events.json","paper":"https://pith.science/paper/SDT6RVDX"},"agent_actions":{"view_html":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7","download_json":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7.json","view_paper":"https://pith.science/paper/SDT6RVDX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.28109&json=true","fetch_graph":"https://pith.science/api/pith-number/SDT6RVDXUNVLK7LVGN2HWPIPX7/graph.json","fetch_events":"https://pith.science/api/pith-number/SDT6RVDXUNVLK7LVGN2HWPIPX7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7/action/storage_attestation","attest_author":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7/action/author_attestation","sign_citation":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7/action/citation_signature","submit_replication":"https://pith.science/pith/SDT6RVDXUNVLK7LVGN2HWPIPX7/action/replication_record"}},"created_at":"2026-05-28T01:04:59.121813+00:00","updated_at":"2026-05-28T01:04:59.121813+00:00"}