{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ELYC5DAX4QJ3BJCC72RXGVGMH2","short_pith_number":"pith:ELYC5DAX","schema_version":"1.0","canonical_sha256":"22f02e8c17e413b0a442fea37354cc3e8955553a48490247ea5a4d3ecb83452d","source":{"kind":"arxiv","id":"2605.25198","version":1},"attestation_state":"computed","paper":{"title":"Hide to Guide: Learning via Semantic Masking","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Alex Hu, Han Cai, Luke J. Huang, Qinghao Hu, Ruitao Liu, Shang Yang, Song Han, Yecheng Wu, Zhuoyang Zhang","submitted_at":"2026-05-24T17:59:06Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has become a powerful paradigm for improving language models on reasoning-intensive tasks, but its effectiveness is often limited by exploration. For example, models often fail on hard problems, leaving little useful reward signal. External expert traces offer a natural source of guidance, yet they may also expose reward-relevant content along the critical path to the verifier target, such as final answers, intermediate values, executable implementations, or answer-related entities. This content can create an unintended reward hacking chann"},"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":"2605.25198","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-24T17:59:06Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"db9ea028910855da3588485d5994cd14861313c7aa0a7875de901c6c2508be7e","abstract_canon_sha256":"2c308a9f50cbf567955292f4033262a6b582ee05857088c1edeeaad22d569e32"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:04:22.817478Z","signature_b64":"r2OOJiOxBjTti/Z2ZGgVlI8SREaf5RS9A2+GGe5OSwMxBw81A5PGLgwvtCI+Rpx5IIsQHQ6E3PSWkl6HsqMpBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"22f02e8c17e413b0a442fea37354cc3e8955553a48490247ea5a4d3ecb83452d","last_reissued_at":"2026-05-26T02:04:22.816935Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:04:22.816935Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hide to Guide: Learning via Semantic Masking","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Alex Hu, Han Cai, Luke J. Huang, Qinghao Hu, Ruitao Liu, Shang Yang, Song Han, Yecheng Wu, Zhuoyang Zhang","submitted_at":"2026-05-24T17:59:06Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has become a powerful paradigm for improving language models on reasoning-intensive tasks, but its effectiveness is often limited by exploration. For example, models often fail on hard problems, leaving little useful reward signal. External expert traces offer a natural source of guidance, yet they may also expose reward-relevant content along the critical path to the verifier target, such as final answers, intermediate values, executable implementations, or answer-related entities. This content can create an unintended reward hacking chann"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.25198","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/2605.25198/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":"2605.25198","created_at":"2026-05-26T02:04:22.817002+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.25198v1","created_at":"2026-05-26T02:04:22.817002+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.25198","created_at":"2026-05-26T02:04:22.817002+00:00"},{"alias_kind":"pith_short_12","alias_value":"ELYC5DAX4QJ3","created_at":"2026-05-26T02:04:22.817002+00:00"},{"alias_kind":"pith_short_16","alias_value":"ELYC5DAX4QJ3BJCC","created_at":"2026-05-26T02:04:22.817002+00:00"},{"alias_kind":"pith_short_8","alias_value":"ELYC5DAX","created_at":"2026-05-26T02:04:22.817002+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/ELYC5DAX4QJ3BJCC72RXGVGMH2","json":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2.json","graph_json":"https://pith.science/api/pith-number/ELYC5DAX4QJ3BJCC72RXGVGMH2/graph.json","events_json":"https://pith.science/api/pith-number/ELYC5DAX4QJ3BJCC72RXGVGMH2/events.json","paper":"https://pith.science/paper/ELYC5DAX"},"agent_actions":{"view_html":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2","download_json":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2.json","view_paper":"https://pith.science/paper/ELYC5DAX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.25198&json=true","fetch_graph":"https://pith.science/api/pith-number/ELYC5DAX4QJ3BJCC72RXGVGMH2/graph.json","fetch_events":"https://pith.science/api/pith-number/ELYC5DAX4QJ3BJCC72RXGVGMH2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2/action/storage_attestation","attest_author":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2/action/author_attestation","sign_citation":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2/action/citation_signature","submit_replication":"https://pith.science/pith/ELYC5DAX4QJ3BJCC72RXGVGMH2/action/replication_record"}},"created_at":"2026-05-26T02:04:22.817002+00:00","updated_at":"2026-05-26T02:04:22.817002+00:00"}