pith:7H7ZXYMA
Relational Preference Encoding in Looped Transformer Internal States
Looped transformers encode human preferences relationally through comparisons in their internal states.
arxiv:2604.09870 v2 · 2026-04-10 · cs.LG · cs.AI
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{7H7ZXYMAS22SQM3SKLMYYUDPM6}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
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.
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.
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.
Formal links
Receipt and verification
| First computed | 2026-07-17T01:21:48.919085Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/7H7ZXYMAS22SQM3SKLMYYUDPM6 \
| jq -c '.canonical_record' \
| python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: f9ff9be18096b528337252d98c506f678a6eeff4f605b22d5e7f8b902c04dac7
Canonical record JSON
{
"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
}
}