{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LLHS5SOIMDCNTQBKJKZ6FCIQFU","short_pith_number":"pith:LLHS5SOI","schema_version":"1.0","canonical_sha256":"5acf2ec9c860c4d9c02a4ab3e289102d0555d593dc15627b811ec8a5bfdce8c6","source":{"kind":"arxiv","id":"2605.28247","version":1},"attestation_state":"computed","paper":{"title":"IRDS: Interpretable RLVR Data Selection via Verifier-Coupled Sparse Autoencoder Coverage","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dazhong Shen, Mingxu Zhang, Ying Sun, Yuhan Li","submitted_at":"2026-05-27T09:58:05Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has become a key technique for en- hancing LLM reasoning, yet its data ineffi- ciency remains a major bottleneck. Existing methods address this problem only partially, each missing at least one of subset-level cov- erage, verifier signal use, or interpretability. To address this gap, we present IRDS (Inter- pretable RLVR Data Selection), which selects RLVR training instances on a sparse autoen- coder (SAE) cluster basis so the selection itself is auditable on recognizable problem motifs. To select instances the model both fails on and can s"},"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":"2605.28247","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T09:58:05Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7722b3ab71c485fb08e96023ddd9cd3dfbf8a058b0fa0bf5d4243716c43520b3","abstract_canon_sha256":"6b32e68a92d4ada65764f756dadce61e18e892aead0c77ef03cb4636de44d6fc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T01:05:03.993224Z","signature_b64":"uG7GhGBgEy4hebas/3oZqtStPmPNRP3INQ8MMM4rZkj5tKmA9Dx2oaSshgUEgw1RNKZKdRpIUVd5Id62BOcrBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5acf2ec9c860c4d9c02a4ab3e289102d0555d593dc15627b811ec8a5bfdce8c6","last_reissued_at":"2026-05-28T01:05:03.992743Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T01:05:03.992743Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"IRDS: Interpretable RLVR Data Selection via Verifier-Coupled Sparse Autoencoder Coverage","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Dazhong Shen, Mingxu Zhang, Ying Sun, Yuhan Li","submitted_at":"2026-05-27T09:58:05Z","abstract_excerpt":"Reinforcement learning with verifiable rewards (RLVR) has become a key technique for en- hancing LLM reasoning, yet its data ineffi- ciency remains a major bottleneck. Existing methods address this problem only partially, each missing at least one of subset-level cov- erage, verifier signal use, or interpretability. To address this gap, we present IRDS (Inter- pretable RLVR Data Selection), which selects RLVR training instances on a sparse autoen- coder (SAE) cluster basis so the selection itself is auditable on recognizable problem motifs. To select instances the model both fails on and can s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.28247","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/2605.28247/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":"2605.28247","created_at":"2026-05-28T01:05:03.992812+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.28247v1","created_at":"2026-05-28T01:05:03.992812+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.28247","created_at":"2026-05-28T01:05:03.992812+00:00"},{"alias_kind":"pith_short_12","alias_value":"LLHS5SOIMDCN","created_at":"2026-05-28T01:05:03.992812+00:00"},{"alias_kind":"pith_short_16","alias_value":"LLHS5SOIMDCNTQBK","created_at":"2026-05-28T01:05:03.992812+00:00"},{"alias_kind":"pith_short_8","alias_value":"LLHS5SOI","created_at":"2026-05-28T01:05:03.992812+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/LLHS5SOIMDCNTQBKJKZ6FCIQFU","json":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU.json","graph_json":"https://pith.science/api/pith-number/LLHS5SOIMDCNTQBKJKZ6FCIQFU/graph.json","events_json":"https://pith.science/api/pith-number/LLHS5SOIMDCNTQBKJKZ6FCIQFU/events.json","paper":"https://pith.science/paper/LLHS5SOI"},"agent_actions":{"view_html":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU","download_json":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU.json","view_paper":"https://pith.science/paper/LLHS5SOI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.28247&json=true","fetch_graph":"https://pith.science/api/pith-number/LLHS5SOIMDCNTQBKJKZ6FCIQFU/graph.json","fetch_events":"https://pith.science/api/pith-number/LLHS5SOIMDCNTQBKJKZ6FCIQFU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU/action/storage_attestation","attest_author":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU/action/author_attestation","sign_citation":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU/action/citation_signature","submit_replication":"https://pith.science/pith/LLHS5SOIMDCNTQBKJKZ6FCIQFU/action/replication_record"}},"created_at":"2026-05-28T01:05:03.992812+00:00","updated_at":"2026-05-28T01:05:03.992812+00:00"}