{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:44AVRDDWDUFQPJS35YWFFIWS27","short_pith_number":"pith:44AVRDDW","schema_version":"1.0","canonical_sha256":"e701588c761d0b07a65bee2c52a2d2d7d61bdbcd8cda9e2bf8769fe6590ca423","source":{"kind":"arxiv","id":"2606.09822","version":1},"attestation_state":"computed","paper":{"title":"Causally Evaluating the Learnability of Formal Language Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.FL"],"primary_cat":"cs.CL","authors_text":"Anej Svete, Brian DuSell, Josef Valvoda, Reda Boumasmoud, Ryan Cotterell, V\\'esteinn Sn{\\ae}bjarnarson","submitted_at":"2026-06-08T17:58:36Z","abstract_excerpt":"Language models, as multi-task learners, acquire a wide range of abilities during training. A fundamental question is how much task-specific data is needed to learn a given task. Answering this for natural language is difficult: tasks are hard to delineate and can confound one another. To rigorously investigate the relationship between data frequency and learnability, we turn to a controlled setting using formal languages induced from probabilistic finite automata. These serve as a methodological testbed to demonstrate that standard correlational evaluation practices are inherently flawed. To "},"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":"2606.09822","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-08T17:58:36Z","cross_cats_sorted":["cs.FL"],"title_canon_sha256":"86be71472acfda16808c58250b3e58249bb43001a0ceaef8f7cb8994eb916efa","abstract_canon_sha256":"86f4237d5faa716f3d75b994fd599ddbae26712a85946b9d3c5b05e8d1113418"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:09:12.491358Z","signature_b64":"k4mcUwU+CREpzc/mmr6ZyZ+c7Ha61I2k8j4ZfZ+v93/wqgKEvEVgRZQlWBpLa/pBanD4M68CY/M9n8XwiBSyBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e701588c761d0b07a65bee2c52a2d2d7d61bdbcd8cda9e2bf8769fe6590ca423","last_reissued_at":"2026-06-09T02:09:12.490514Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:09:12.490514Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Causally Evaluating the Learnability of Formal Language Tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.FL"],"primary_cat":"cs.CL","authors_text":"Anej Svete, Brian DuSell, Josef Valvoda, Reda Boumasmoud, Ryan Cotterell, V\\'esteinn Sn{\\ae}bjarnarson","submitted_at":"2026-06-08T17:58:36Z","abstract_excerpt":"Language models, as multi-task learners, acquire a wide range of abilities during training. A fundamental question is how much task-specific data is needed to learn a given task. Answering this for natural language is difficult: tasks are hard to delineate and can confound one another. To rigorously investigate the relationship between data frequency and learnability, we turn to a controlled setting using formal languages induced from probabilistic finite automata. These serve as a methodological testbed to demonstrate that standard correlational evaluation practices are inherently flawed. To "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09822","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/2606.09822/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":"2606.09822","created_at":"2026-06-09T02:09:12.490651+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.09822v1","created_at":"2026-06-09T02:09:12.490651+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09822","created_at":"2026-06-09T02:09:12.490651+00:00"},{"alias_kind":"pith_short_12","alias_value":"44AVRDDWDUFQ","created_at":"2026-06-09T02:09:12.490651+00:00"},{"alias_kind":"pith_short_16","alias_value":"44AVRDDWDUFQPJS3","created_at":"2026-06-09T02:09:12.490651+00:00"},{"alias_kind":"pith_short_8","alias_value":"44AVRDDW","created_at":"2026-06-09T02:09:12.490651+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/44AVRDDWDUFQPJS35YWFFIWS27","json":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27.json","graph_json":"https://pith.science/api/pith-number/44AVRDDWDUFQPJS35YWFFIWS27/graph.json","events_json":"https://pith.science/api/pith-number/44AVRDDWDUFQPJS35YWFFIWS27/events.json","paper":"https://pith.science/paper/44AVRDDW"},"agent_actions":{"view_html":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27","download_json":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27.json","view_paper":"https://pith.science/paper/44AVRDDW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.09822&json=true","fetch_graph":"https://pith.science/api/pith-number/44AVRDDWDUFQPJS35YWFFIWS27/graph.json","fetch_events":"https://pith.science/api/pith-number/44AVRDDWDUFQPJS35YWFFIWS27/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27/action/timestamp_anchor","attest_storage":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27/action/storage_attestation","attest_author":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27/action/author_attestation","sign_citation":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27/action/citation_signature","submit_replication":"https://pith.science/pith/44AVRDDWDUFQPJS35YWFFIWS27/action/replication_record"}},"created_at":"2026-06-09T02:09:12.490651+00:00","updated_at":"2026-06-09T02:09:12.490651+00:00"}