{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3LZRQ3IJSOTTSOYZEEP5TU7WSE","short_pith_number":"pith:3LZRQ3IJ","schema_version":"1.0","canonical_sha256":"daf3186d0993a7393b19211fd9d3f691361d7dc526862d976e0fd26a04be7061","source":{"kind":"arxiv","id":"2501.02266","version":1},"attestation_state":"computed","paper":{"title":"LLMzSz{\\L}: a comprehensive LLM benchmark for Polish","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Filip Grali\\'nski, Jakub Pokrywka, Krzysztof Jassem, Marek Kubis, Micha{\\l} Ciesi\\'o{\\l}ka, Monika Jab{\\l}o\\'nska, Piotr Jab{\\l}o\\'nski, Ryszard Staruch","submitted_at":"2025-01-04T12:04:46Z","abstract_excerpt":"This article introduces the first comprehensive benchmark for the Polish language at this scale: LLMzSz{\\L} (LLMs Behind the School Desk). It is based on a coherent collection of Polish national exams, including both academic and professional tests extracted from the archives of the Polish Central Examination Board. It covers 4 types of exams, coming from 154 domains. Altogether, it consists of almost 19k closed-ended questions. We investigate the performance of open-source multilingual, English, and Polish LLMs to verify LLMs' abilities to transfer knowledge between languages. Also, the corre"},"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":"2501.02266","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-01-04T12:04:46Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1b1474366ef8bba0de60384768a8d507a4700fbbca02d16f46503ed4fad67378","abstract_canon_sha256":"b5d9f5518e89229c83c3020aaffb60f806c0e6998804d04ccdb28381eee21ab0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:57:05.062960Z","signature_b64":"LP3ZwH/VUc2atFdpb9Fgp5GW3Ao61fB32mUAeDNHjCglsYRE/Brpq9smxoeNQI+RZ97YlDyOh4UWm6G0ZkpbAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"daf3186d0993a7393b19211fd9d3f691361d7dc526862d976e0fd26a04be7061","last_reissued_at":"2026-07-05T09:57:05.062608Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:57:05.062608Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLMzSz{\\L}: a comprehensive LLM benchmark for Polish","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Filip Grali\\'nski, Jakub Pokrywka, Krzysztof Jassem, Marek Kubis, Micha{\\l} Ciesi\\'o{\\l}ka, Monika Jab{\\l}o\\'nska, Piotr Jab{\\l}o\\'nski, Ryszard Staruch","submitted_at":"2025-01-04T12:04:46Z","abstract_excerpt":"This article introduces the first comprehensive benchmark for the Polish language at this scale: LLMzSz{\\L} (LLMs Behind the School Desk). It is based on a coherent collection of Polish national exams, including both academic and professional tests extracted from the archives of the Polish Central Examination Board. It covers 4 types of exams, coming from 154 domains. Altogether, it consists of almost 19k closed-ended questions. We investigate the performance of open-source multilingual, English, and Polish LLMs to verify LLMs' abilities to transfer knowledge between languages. Also, the corre"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.02266","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/2501.02266/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":"2501.02266","created_at":"2026-07-05T09:57:05.062667+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.02266v1","created_at":"2026-07-05T09:57:05.062667+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.02266","created_at":"2026-07-05T09:57:05.062667+00:00"},{"alias_kind":"pith_short_12","alias_value":"3LZRQ3IJSOTT","created_at":"2026-07-05T09:57:05.062667+00:00"},{"alias_kind":"pith_short_16","alias_value":"3LZRQ3IJSOTTSOYZ","created_at":"2026-07-05T09:57:05.062667+00:00"},{"alias_kind":"pith_short_8","alias_value":"3LZRQ3IJ","created_at":"2026-07-05T09:57:05.062667+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.16322","citing_title":"PL-Guard: Benchmarking Language Model Safety for Polish","ref_index":10,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE","json":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE.json","graph_json":"https://pith.science/api/pith-number/3LZRQ3IJSOTTSOYZEEP5TU7WSE/graph.json","events_json":"https://pith.science/api/pith-number/3LZRQ3IJSOTTSOYZEEP5TU7WSE/events.json","paper":"https://pith.science/paper/3LZRQ3IJ"},"agent_actions":{"view_html":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE","download_json":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE.json","view_paper":"https://pith.science/paper/3LZRQ3IJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.02266&json=true","fetch_graph":"https://pith.science/api/pith-number/3LZRQ3IJSOTTSOYZEEP5TU7WSE/graph.json","fetch_events":"https://pith.science/api/pith-number/3LZRQ3IJSOTTSOYZEEP5TU7WSE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE/action/storage_attestation","attest_author":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE/action/author_attestation","sign_citation":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE/action/citation_signature","submit_replication":"https://pith.science/pith/3LZRQ3IJSOTTSOYZEEP5TU7WSE/action/replication_record"}},"created_at":"2026-07-05T09:57:05.062667+00:00","updated_at":"2026-07-05T09:57:05.062667+00:00"}