{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Z7V6WVBZRUGKHRB6ISE6REE5RG","short_pith_number":"pith:Z7V6WVBZ","schema_version":"1.0","canonical_sha256":"cfebeb54398d0ca3c43e4489e8909d899f22e4a6442c561c5a1d8edad34f84b7","source":{"kind":"arxiv","id":"2607.14427","version":1},"attestation_state":"computed","paper":{"title":"Per-Token Fixed-Point Convergence in Depth-Recurrent Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Joe Logan","submitted_at":"2026-07-15T23:38:44Z","abstract_excerpt":"A depth-recurrent transformer applies a weight-tied core a variable number of times, and prior work has shown that training with a randomized recursion count yields one checkpoint usable across a range of inference depths. We ask what such a model actually computes per token, and measure it directly. On a 135M-class model trained on FineWeb-Edu, the recurrent state converges to a per-token fixed point: mean successive-output KL divergence falls from 3.9e-1 at the second loop to 8.5e-6 by the sixteenth, and per-token state change decays in step. Crucially, this convergence is not uniform across"},"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":"2607.14427","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-07-15T23:38:44Z","cross_cats_sorted":[],"title_canon_sha256":"adfdad9fe1810c6386d5b013ad2be0168561bc264c061d4b68f194f7b26051ba","abstract_canon_sha256":"ee66b129d1016f57ebfdf107be960c2a1d37b38df6b7c5efc98f43b6427e5e8d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-17T00:21:11.896619Z","signature_b64":"ZX4bv1++2qF1Wux/Y85VbQcQaEA5zRDwsfyUaduw6KBH1Svmea5yJ2zpqDTfkGcUfgSCxE+kq2xjq1UW3ELBCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cfebeb54398d0ca3c43e4489e8909d899f22e4a6442c561c5a1d8edad34f84b7","last_reissued_at":"2026-07-17T00:21:11.895774Z","signature_status":"signed_v1","first_computed_at":"2026-07-17T00:21:11.895774Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Per-Token Fixed-Point Convergence in Depth-Recurrent Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Joe Logan","submitted_at":"2026-07-15T23:38:44Z","abstract_excerpt":"A depth-recurrent transformer applies a weight-tied core a variable number of times, and prior work has shown that training with a randomized recursion count yields one checkpoint usable across a range of inference depths. We ask what such a model actually computes per token, and measure it directly. On a 135M-class model trained on FineWeb-Edu, the recurrent state converges to a per-token fixed point: mean successive-output KL divergence falls from 3.9e-1 at the second loop to 8.5e-6 by the sixteenth, and per-token state change decays in step. Crucially, this convergence is not uniform across"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.14427","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/2607.14427/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":"2607.14427","created_at":"2026-07-17T00:21:11.896206+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.14427v1","created_at":"2026-07-17T00:21:11.896206+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.14427","created_at":"2026-07-17T00:21:11.896206+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z7V6WVBZRUGK","created_at":"2026-07-17T00:21:11.896206+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z7V6WVBZRUGKHRB6","created_at":"2026-07-17T00:21:11.896206+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z7V6WVBZ","created_at":"2026-07-17T00:21:11.896206+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/Z7V6WVBZRUGKHRB6ISE6REE5RG","json":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG.json","graph_json":"https://pith.science/api/pith-number/Z7V6WVBZRUGKHRB6ISE6REE5RG/graph.json","events_json":"https://pith.science/api/pith-number/Z7V6WVBZRUGKHRB6ISE6REE5RG/events.json","paper":"https://pith.science/paper/Z7V6WVBZ"},"agent_actions":{"view_html":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG","download_json":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG.json","view_paper":"https://pith.science/paper/Z7V6WVBZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.14427&json=true","fetch_graph":"https://pith.science/api/pith-number/Z7V6WVBZRUGKHRB6ISE6REE5RG/graph.json","fetch_events":"https://pith.science/api/pith-number/Z7V6WVBZRUGKHRB6ISE6REE5RG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG/action/storage_attestation","attest_author":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG/action/author_attestation","sign_citation":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG/action/citation_signature","submit_replication":"https://pith.science/pith/Z7V6WVBZRUGKHRB6ISE6REE5RG/action/replication_record"}},"created_at":"2026-07-17T00:21:11.896206+00:00","updated_at":"2026-07-17T00:21:11.896206+00:00"}