{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KKK6GEDMK7KL4SCRFFZLBAIQK3","short_pith_number":"pith:KKK6GEDM","schema_version":"1.0","canonical_sha256":"5295e3106c57d4be48512972b0811056ce21cb533129953a24245ffc3287cc27","source":{"kind":"arxiv","id":"2605.29888","version":1},"attestation_state":"computed","paper":{"title":"LaRA: Layer-wise Representation Analysis for Detecting Data Contamination in RL Post-Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Alan Ritter, Dongseok Lee, Guijin Son, Jaehyung Kim, Minju Gwak, Minseo Kwak","submitted_at":"2026-05-28T13:13:49Z","abstract_excerpt":"Reinforcement learning (RL) post-training has shown to improve reasoning in large language models (LLMs). However, there has been little exploration on the problem of data contamination in RL post-training, potentially undermining generalization and evaluation reliability of the training process itself. Existing detection methods primarily rely on output-level signals such as likelihood or entropy, which become unreliable for RL-trained models since RL shapes behavior through trajectory-level rewards rather than token likelihoods. We propose LaRA, a layer-wise representation analysis framework"},"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":"2605.29888","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T13:13:49Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2e21f2045ce0ec770d8a252044d4ff7a8ad55eed53a6987b864945111fdf7cf9","abstract_canon_sha256":"f87c17dbec146f48ab925260964d26365a8e76b9dd4deeb269fc9f2835c73b47"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:05:57.884798Z","signature_b64":"8GWP3PZNb51JRcPEuzHnSYcMnsF6AE/GpB4HbVPWXbUuR6I5yqAWxKwdHXg5rs31OlRoNVBBEOuI0ttNarqICw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5295e3106c57d4be48512972b0811056ce21cb533129953a24245ffc3287cc27","last_reissued_at":"2026-05-29T02:05:57.884034Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:05:57.884034Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LaRA: Layer-wise Representation Analysis for Detecting Data Contamination in RL Post-Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Alan Ritter, Dongseok Lee, Guijin Son, Jaehyung Kim, Minju Gwak, Minseo Kwak","submitted_at":"2026-05-28T13:13:49Z","abstract_excerpt":"Reinforcement learning (RL) post-training has shown to improve reasoning in large language models (LLMs). However, there has been little exploration on the problem of data contamination in RL post-training, potentially undermining generalization and evaluation reliability of the training process itself. Existing detection methods primarily rely on output-level signals such as likelihood or entropy, which become unreliable for RL-trained models since RL shapes behavior through trajectory-level rewards rather than token likelihoods. We propose LaRA, a layer-wise representation analysis framework"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.29888","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/2605.29888/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":"2605.29888","created_at":"2026-05-29T02:05:57.884149+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.29888v1","created_at":"2026-05-29T02:05:57.884149+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.29888","created_at":"2026-05-29T02:05:57.884149+00:00"},{"alias_kind":"pith_short_12","alias_value":"KKK6GEDMK7KL","created_at":"2026-05-29T02:05:57.884149+00:00"},{"alias_kind":"pith_short_16","alias_value":"KKK6GEDMK7KL4SCR","created_at":"2026-05-29T02:05:57.884149+00:00"},{"alias_kind":"pith_short_8","alias_value":"KKK6GEDM","created_at":"2026-05-29T02:05:57.884149+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/KKK6GEDMK7KL4SCRFFZLBAIQK3","json":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3.json","graph_json":"https://pith.science/api/pith-number/KKK6GEDMK7KL4SCRFFZLBAIQK3/graph.json","events_json":"https://pith.science/api/pith-number/KKK6GEDMK7KL4SCRFFZLBAIQK3/events.json","paper":"https://pith.science/paper/KKK6GEDM"},"agent_actions":{"view_html":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3","download_json":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3.json","view_paper":"https://pith.science/paper/KKK6GEDM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.29888&json=true","fetch_graph":"https://pith.science/api/pith-number/KKK6GEDMK7KL4SCRFFZLBAIQK3/graph.json","fetch_events":"https://pith.science/api/pith-number/KKK6GEDMK7KL4SCRFFZLBAIQK3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3/action/storage_attestation","attest_author":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3/action/author_attestation","sign_citation":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3/action/citation_signature","submit_replication":"https://pith.science/pith/KKK6GEDMK7KL4SCRFFZLBAIQK3/action/replication_record"}},"created_at":"2026-05-29T02:05:57.884149+00:00","updated_at":"2026-05-29T02:05:57.884149+00:00"}