{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:7H7ZXYMAS22SQM3SKLMYYUDPM6","short_pith_number":"pith:7H7ZXYMA","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"},"canonical_sha256":"f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7","source":{"kind":"arxiv","id":"2604.09870","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.09870","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"arxiv_version","alias_value":"2604.09870v2","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.09870","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"pith_short_12","alias_value":"7H7ZXYMAS22S","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"pith_short_16","alias_value":"7H7ZXYMAS22SQM3S","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"pith_short_8","alias_value":"7H7ZXYMA","created_at":"2026-07-17T01:21:48Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:7H7ZXYMAS22SQM3SKLMYYUDPM6","target":"record","payload":{"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"},"canonical_sha256":"f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7","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"},"source_kind":"arxiv","source_id":"2604.09870","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-17T01:21:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"k6AKfnVGaCkJW6cnw9+lsG+w0MxtkawgzrQlCjibvLHMlHINJVipyd/GPipch7tRe1L4dyBy6d4on3rgdjtbAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T17:03:58.991427Z"},"content_sha256":"864e08f69fa5f683e998b865fcabfbe3fd30c843948bc393c3d6db5959eeda92","schema_version":"1.0","event_id":"sha256:864e08f69fa5f683e998b865fcabfbe3fd30c843948bc393c3d6db5959eeda92"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:7H7ZXYMAS22SQM3SKLMYYUDPM6","target":"graph","payload":{"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"},"verdict_id":"4bf0a652-a8db-4c35-9aa3-faebbfe6d3f5"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-17T01:21:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"qUFtBmtNR/n+mcRQxqa1cLltpBvgMbnVy9c3X4eTh8BHAzkBBW6J8u3w9kEc3sG+bXwwk2sKB5jG5WEvG77lDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T17:03:58.992174Z"},"content_sha256":"ad204e0be21c8672390ee6d9eb5a4d58429bb56e48f5d8b128dea415e0a6e168","schema_version":"1.0","event_id":"sha256:ad204e0be21c8672390ee6d9eb5a4d58429bb56e48f5d8b128dea415e0a6e168"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/bundle.json","state_url":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-06T17:03:58Z","links":{"resolver":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6","bundle":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/bundle.json","state":"https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/state.json","well_known_bundle":"https://pith.science/.well-known/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:7H7ZXYMAS22SQM3SKLMYYUDPM6","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"dbc00ddc9a50c56343ccc7816f66cf17e6d006713f17c1726c2463b5a7c07d73","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-10T20:00:49Z","title_canon_sha256":"d35755bea99d8c5205a3b7b40e76c221753ce8b234cca850a1ce6048eb9f6b8d"},"schema_version":"1.0","source":{"id":"2604.09870","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.09870","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"arxiv_version","alias_value":"2604.09870v2","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.09870","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"pith_short_12","alias_value":"7H7ZXYMAS22S","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"pith_short_16","alias_value":"7H7ZXYMAS22SQM3S","created_at":"2026-07-17T01:21:48Z"},{"alias_kind":"pith_short_8","alias_value":"7H7ZXYMA","created_at":"2026-07-17T01:21:48Z"}],"graph_snapshots":[{"event_id":"sha256:ad204e0be21c8672390ee6d9eb5a4d58429bb56e48f5d8b128dea415e0a6e168","target":"graph","created_at":"2026-07-17T01:21:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Looped transformers encode human preferences relationally through comparisons in their internal states."}],"snapshot_sha256":"5d2973ffe2c842584fa0e45d21b06c64425c9a0eae02bef254fc29f714ecd0ef"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"079fbfd2c937727fbd75265f07da373d1df5b5a423a446a52f1a98c7d2ddddac"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2604.09870/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"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","authors_text":"Jan Kirin","cross_cats":["cs.AI"],"headline":"Looped transformers encode human preferences relationally through comparisons in their internal states.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-10T20:00:49Z","title":"Relational Preference Encoding in Looped Transformer Internal States"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.09870","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-10T16:53:36.083854Z","id":"4bf0a652-a8db-4c35-9aa3-faebbfe6d3f5","model_set":{"reader":"grok-4.3"},"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","pith_extraction_headline":"Looped transformers encode human preferences relationally through comparisons in their internal states.","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.","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."}},"verdict_id":"4bf0a652-a8db-4c35-9aa3-faebbfe6d3f5"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:864e08f69fa5f683e998b865fcabfbe3fd30c843948bc393c3d6db5959eeda92","target":"record","created_at":"2026-07-17T01:21:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"dbc00ddc9a50c56343ccc7816f66cf17e6d006713f17c1726c2463b5a7c07d73","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-04-10T20:00:49Z","title_canon_sha256":"d35755bea99d8c5205a3b7b40e76c221753ce8b234cca850a1ce6048eb9f6b8d"},"schema_version":"1.0","source":{"id":"2604.09870","kind":"arxiv","version":2}},"canonical_sha256":"f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7","first_computed_at":"2026-07-17T01:21:48.919085Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-17T01:21:48.919085Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"uwTQ9yC8e6q5Ai++Ncbla7TRIx5uM9kJyU8uQH32HO21hfPGaWyJPzFBbqeZI2x2Ldx0UNdgQMA+x3N2YCDjBA==","signature_status":"signed_v1","signed_at":"2026-07-17T01:21:48.919961Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.09870","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:864e08f69fa5f683e998b865fcabfbe3fd30c843948bc393c3d6db5959eeda92","sha256:ad204e0be21c8672390ee6d9eb5a4d58429bb56e48f5d8b128dea415e0a6e168"],"state_sha256":"b00c7afc7b853d280a6ebb48ffd4cbe6d37d4e77af0b904ba5e5608c452e0a98"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UQB6vEmnLbkG7ygmbM7sH/WdqHhJ8JraSWl/KSwaSN+054V+0GG50r5AnsuSiJeV9RPGwKIEFB159G3P4zcTDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T17:03:58.996638Z","bundle_sha256":"a0951b7e1ba85074fd7cfe52acec00c633da17f09dfd5027fd45e434df449af8"}}