{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7H7ZXYMAS22SQM3SKLMYYUDPM6","short_pith_number":"pith:7H7ZXYMA","schema_version":"1.0","canonical_sha256":"f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7","source":{"kind":"arxiv","id":"2604.09870","version":2},"attestation_state":"computed","paper":{"title":"Relational Preference Encoding in Looped Transformer Internal States","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Looped transformers encode human preferences relationally through comparisons in their internal states.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jan Kirin","submitted_at":"2026-04-10T20:00:49Z","abstract_excerpt":"We investigate how looped transformers encode human preference, training lightweight evaluator heads on frozen Ouro-2.6B loop-iteration states on Anthropic HH-RLHF.\n  v2: an erratum is prepended; the original manuscript is unchanged. A post-publication audit found the three headline results inflated by two independent evaluation errors. The 95.2% pairwise evaluator accuracy is a canonical-ordering artifact: the data were correctly split, but the evaluator learned to prefer the first-presented argument; its strict antisymmetrized accuracy on the full 8,552-pair test set is 63.9%. The 84.5% pair"},"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":true},"canonical_record":{"source":{"id":"2604.09870","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-10T20:00:49Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d35755bea99d8c5205a3b7b40e76c221753ce8b234cca850a1ce6048eb9f6b8d","abstract_canon_sha256":"dbc00ddc9a50c56343ccc7816f66cf17e6d006713f17c1726c2463b5a7c07d73"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-17T01:21:48.919961Z","signature_b64":"uwTQ9yC8e6q5Ai++Ncbla7TRIx5uM9kJyU8uQH32HO21hfPGaWyJPzFBbqeZI2x2Ldx0UNdgQMA+x3N2YCDjBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7","last_reissued_at":"2026-07-17T01:21:48.919085Z","signature_status":"signed_v1","first_computed_at":"2026-07-17T01:21:48.919085Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Relational Preference Encoding in Looped Transformer Internal States","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Looped transformers encode human preferences relationally through comparisons in their internal states.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jan Kirin","submitted_at":"2026-04-10T20:00:49Z","abstract_excerpt":"We investigate how looped transformers encode human preference, training lightweight evaluator heads on frozen Ouro-2.6B loop-iteration states on Anthropic HH-RLHF.\n  v2: an erratum is prepended; the original manuscript is unchanged. A post-publication audit found the three headline results inflated by two independent evaluation errors. The 95.2% pairwise evaluator accuracy is a canonical-ordering artifact: the data were correctly split, but the evaluator learned to prefer the first-presented argument; its strict antisymmetrized accuracy on the full 8,552-pair test set is 63.9%. The 84.5% pair"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"loop states encode preference predominantly relationally: a linear probe on pairwise differences achieves 84.5%, the best nonlinear independent evaluator reaches only 65% test accuracy, and linear independent classification scores 21.75%, below chance and with inverted polarity.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the high pairwise accuracy and low independent accuracy reflect genuine relational encoding in the model's learned value system rather than an artifact of the 50% argument-swap protocol, the accidental early stopping, or the specific choice of lightweight heads trained on the same model's representations.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Looped transformer hidden states encode preferences relationally via pairwise differences rather than independent pointwise classification, with the evaluator acting as an internal consistency probe on the model's own value system.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Looped transformers encode human preferences relationally through comparisons in their internal states.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5d2973ffe2c842584fa0e45d21b06c64425c9a0eae02bef254fc29f714ecd0ef"},"source":{"id":"2604.09870","kind":"arxiv","version":2},"verdict":{"id":"4bf0a652-a8db-4c35-9aa3-faebbfe6d3f5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T16:53:36.083854Z","strongest_claim":"loop states encode preference predominantly relationally: a linear probe on pairwise differences achieves 84.5%, the best nonlinear independent evaluator reaches only 65% test accuracy, and linear independent classification scores 21.75%, below chance and with inverted polarity.","one_line_summary":"Looped transformer hidden states encode preferences relationally via pairwise differences rather than independent pointwise classification, with the evaluator acting as an internal consistency probe on the model's own value system.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the high pairwise accuracy and low independent accuracy reflect genuine relational encoding in the model's learned value system rather than an artifact of the 50% argument-swap protocol, the accidental early stopping, or the specific choice of lightweight heads trained on the same model's representations.","pith_extraction_headline":"Looped transformers encode human preferences relationally through comparisons in their internal states."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.09870/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":2,"snapshot_sha256":"079fbfd2c937727fbd75265f07da373d1df5b5a423a446a52f1a98c7d2ddddac"},"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":"2604.09870","created_at":"2026-07-17T01:21:48.919521+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.09870v2","created_at":"2026-07-17T01:21:48.919521+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.09870","created_at":"2026-07-17T01:21:48.919521+00:00"},{"alias_kind":"pith_short_12","alias_value":"7H7ZXYMAS22S","created_at":"2026-07-17T01:21:48.919521+00:00"},{"alias_kind":"pith_short_16","alias_value":"7H7ZXYMAS22SQM3S","created_at":"2026-07-17T01:21:48.919521+00:00"},{"alias_kind":"pith_short_8","alias_value":"7H7ZXYMA","created_at":"2026-07-17T01:21:48.919521+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6","json":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6.json","graph_json":"https://pith.science/api/pith-number/7H7ZXYMAS22SQM3SKLMYYUDPM6/graph.json","events_json":"https://pith.science/api/pith-number/7H7ZXYMAS22SQM3SKLMYYUDPM6/events.json","paper":"https://pith.science/paper/7H7ZXYMA"},"agent_actions":{"view_html":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6","download_json":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6.json","view_paper":"https://pith.science/paper/7H7ZXYMA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.09870&json=true","fetch_graph":"https://pith.science/api/pith-number/7H7ZXYMAS22SQM3SKLMYYUDPM6/graph.json","fetch_events":"https://pith.science/api/pith-number/7H7ZXYMAS22SQM3SKLMYYUDPM6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/action/storage_attestation","attest_author":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/action/author_attestation","sign_citation":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/action/citation_signature","submit_replication":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/action/replication_record"}},"created_at":"2026-07-17T01:21:48.919521+00:00","updated_at":"2026-07-17T01:21:48.919521+00:00"}