{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IWBAFUTYAOVQLTIQXI45DUTKRI","short_pith_number":"pith:IWBAFUTY","schema_version":"1.0","canonical_sha256":"458202d27803ab05cd10ba39d1d26a8a06f8d1abc470237e7d80f36011880cc0","source":{"kind":"arxiv","id":"2605.29310","version":1},"attestation_state":"computed","paper":{"title":"Rubric-Guided Process Reward for Stepwise Model Routing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Jian Yang, Shenghao Ye, Shuangwu Chen, Yu Guo, Zhengheng Li","submitted_at":"2026-05-28T03:42:24Z","abstract_excerpt":"Stepwise model routing improves the efficiency of Large Reasoning Models (LRMs) by assigning each reasoning step to a suitable model. Recent methods formulate routing as a sequential decision process and train the router with reinforcement learning. However, although they model routing as a process, they still supervise the router with outcome rewards. Such rewards only reflect final answer correctness and fail to evaluate intermediate routing decisions, which can weaken performance and generalization. To address this gap, we propose RoRo, a rubric-guided process reward framework for stepwise "},"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":"2605.29310","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-28T03:42:24Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"b1478a3656237dcef7e2eebf02abc394599beb19f010550aab746df531e6c976","abstract_canon_sha256":"2ab02a9968cac80752d31ea01e90bed81fcf97b8fefd1e8f6441b04c260b0561"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T01:05:31.975290Z","signature_b64":"yLrcnRnt+ls1bMlTSWwsRrh7sLRa2AKkTmB66LyGXqWe612J7wzUENwVG325PGC+sA8Jc9Q396kiJEGQecz0Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"458202d27803ab05cd10ba39d1d26a8a06f8d1abc470237e7d80f36011880cc0","last_reissued_at":"2026-05-29T01:05:31.974559Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T01:05:31.974559Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rubric-Guided Process Reward for Stepwise Model Routing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Jian Yang, Shenghao Ye, Shuangwu Chen, Yu Guo, Zhengheng Li","submitted_at":"2026-05-28T03:42:24Z","abstract_excerpt":"Stepwise model routing improves the efficiency of Large Reasoning Models (LRMs) by assigning each reasoning step to a suitable model. Recent methods formulate routing as a sequential decision process and train the router with reinforcement learning. However, although they model routing as a process, they still supervise the router with outcome rewards. Such rewards only reflect final answer correctness and fail to evaluate intermediate routing decisions, which can weaken performance and generalization. To address this gap, we propose RoRo, a rubric-guided process reward framework for stepwise "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.29310","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/2605.29310/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":"2605.29310","created_at":"2026-05-29T01:05:31.974671+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.29310v1","created_at":"2026-05-29T01:05:31.974671+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.29310","created_at":"2026-05-29T01:05:31.974671+00:00"},{"alias_kind":"pith_short_12","alias_value":"IWBAFUTYAOVQ","created_at":"2026-05-29T01:05:31.974671+00:00"},{"alias_kind":"pith_short_16","alias_value":"IWBAFUTYAOVQLTIQ","created_at":"2026-05-29T01:05:31.974671+00:00"},{"alias_kind":"pith_short_8","alias_value":"IWBAFUTY","created_at":"2026-05-29T01:05:31.974671+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/IWBAFUTYAOVQLTIQXI45DUTKRI","json":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI.json","graph_json":"https://pith.science/api/pith-number/IWBAFUTYAOVQLTIQXI45DUTKRI/graph.json","events_json":"https://pith.science/api/pith-number/IWBAFUTYAOVQLTIQXI45DUTKRI/events.json","paper":"https://pith.science/paper/IWBAFUTY"},"agent_actions":{"view_html":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI","download_json":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI.json","view_paper":"https://pith.science/paper/IWBAFUTY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.29310&json=true","fetch_graph":"https://pith.science/api/pith-number/IWBAFUTYAOVQLTIQXI45DUTKRI/graph.json","fetch_events":"https://pith.science/api/pith-number/IWBAFUTYAOVQLTIQXI45DUTKRI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI/action/storage_attestation","attest_author":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI/action/author_attestation","sign_citation":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI/action/citation_signature","submit_replication":"https://pith.science/pith/IWBAFUTYAOVQLTIQXI45DUTKRI/action/replication_record"}},"created_at":"2026-05-29T01:05:31.974671+00:00","updated_at":"2026-05-29T01:05:31.974671+00:00"}