{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IWUUTII7ZN5DZZI274CREMWTS5","short_pith_number":"pith:IWUUTII7","schema_version":"1.0","canonical_sha256":"45a949a11fcb7a3ce51aff051232d3974b899cb3268cd79e1107685eaaa76591","source":{"kind":"arxiv","id":"2605.24585","version":1},"attestation_state":"computed","paper":{"title":"Word Class Representations Spontaneously Emerge from Successor Representations Trained on Natural Language","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["q-bio.NC"],"primary_cat":"cs.CL","authors_text":"Achim Schilling, Mathis Immertreu, Patrick Krauss, Thomas Kinfe","submitted_at":"2026-05-23T13:55:09Z","abstract_excerpt":"Language models are typically trained to predict the next token in a sequence. Here, we explore an alternative predictive principle from reinforcement learning: Successor Representations (SRs), which model the expected discounted distribution of future states rather than the immediate next state. We transfer this framework to natural language and train neural networks to predict future word distributions across multiple temporal horizons, thereby learning representations of long-range transition structure. We train a deep residual neural network on WikiText-103 (103 million tokens; 20,000-word"},"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":"2605.24585","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-23T13:55:09Z","cross_cats_sorted":["q-bio.NC"],"title_canon_sha256":"8b418713365f4c65f1ba99fa4d4ee1e985a6bac82f1066e33b51321fe9452d3a","abstract_canon_sha256":"327c6930500bfe4ddb14b9d36bcfecad82c9863279ee4c6aba88e72871845110"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:03:47.568943Z","signature_b64":"VK+mbw0H0iaqZqRgV+HDY1Hmc9ZdxAthlKaHW6KoloqWuQAaGPRzYHQ33F3rkZpEi0nvAl11vBJ06MbZsDRiAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"45a949a11fcb7a3ce51aff051232d3974b899cb3268cd79e1107685eaaa76591","last_reissued_at":"2026-05-26T01:03:47.568178Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:03:47.568178Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Word Class Representations Spontaneously Emerge from Successor Representations Trained on Natural Language","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["q-bio.NC"],"primary_cat":"cs.CL","authors_text":"Achim Schilling, Mathis Immertreu, Patrick Krauss, Thomas Kinfe","submitted_at":"2026-05-23T13:55:09Z","abstract_excerpt":"Language models are typically trained to predict the next token in a sequence. Here, we explore an alternative predictive principle from reinforcement learning: Successor Representations (SRs), which model the expected discounted distribution of future states rather than the immediate next state. We transfer this framework to natural language and train neural networks to predict future word distributions across multiple temporal horizons, thereby learning representations of long-range transition structure. We train a deep residual neural network on WikiText-103 (103 million tokens; 20,000-word"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.24585","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/2605.24585/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":"2605.24585","created_at":"2026-05-26T01:03:47.568304+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.24585v1","created_at":"2026-05-26T01:03:47.568304+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.24585","created_at":"2026-05-26T01:03:47.568304+00:00"},{"alias_kind":"pith_short_12","alias_value":"IWUUTII7ZN5D","created_at":"2026-05-26T01:03:47.568304+00:00"},{"alias_kind":"pith_short_16","alias_value":"IWUUTII7ZN5DZZI2","created_at":"2026-05-26T01:03:47.568304+00:00"},{"alias_kind":"pith_short_8","alias_value":"IWUUTII7","created_at":"2026-05-26T01:03:47.568304+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/IWUUTII7ZN5DZZI274CREMWTS5","json":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5.json","graph_json":"https://pith.science/api/pith-number/IWUUTII7ZN5DZZI274CREMWTS5/graph.json","events_json":"https://pith.science/api/pith-number/IWUUTII7ZN5DZZI274CREMWTS5/events.json","paper":"https://pith.science/paper/IWUUTII7"},"agent_actions":{"view_html":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5","download_json":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5.json","view_paper":"https://pith.science/paper/IWUUTII7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.24585&json=true","fetch_graph":"https://pith.science/api/pith-number/IWUUTII7ZN5DZZI274CREMWTS5/graph.json","fetch_events":"https://pith.science/api/pith-number/IWUUTII7ZN5DZZI274CREMWTS5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5/action/storage_attestation","attest_author":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5/action/author_attestation","sign_citation":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5/action/citation_signature","submit_replication":"https://pith.science/pith/IWUUTII7ZN5DZZI274CREMWTS5/action/replication_record"}},"created_at":"2026-05-26T01:03:47.568304+00:00","updated_at":"2026-05-26T01:03:47.568304+00:00"}