{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LYKD3BVEKXITP7PEHBZ4W7EXNL","short_pith_number":"pith:LYKD3BVE","schema_version":"1.0","canonical_sha256":"5e143d86a455d137fde43873cb7c976afbfd9e3bf27bd743a5877c9abbc3a47c","source":{"kind":"arxiv","id":"2505.17082","version":1},"attestation_state":"computed","paper":{"title":"GemMaroc: Unlocking Darija Proficiency in LLMs with Minimal Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Abderrahman Skiredj, Ferdaous Azhari, Houdaifa Atou, Ismail Berrada, Nouamane Tazi","submitted_at":"2025-05-20T12:38:42Z","abstract_excerpt":"Open-source large language models (LLMs) still marginalise Moroccan Arabic (Darija), forcing practitioners either to bolt on heavyweight Arabic adapters or to sacrifice the very reasoning skills that make LLMs useful. We show that a rigorously quality-over-quantity alignment strategy can surface fluent Darija while safeguarding the backbone s cross-lingual reasoning at a sliver of the usual compute. We translate three compact instruction suites LIMA 1 K, DEITA 6 K and TULU 50 K into Darija, preserve 20 of the English originals, and add mathematics, coding and scientific prompts. A LoRA-tuned G"},"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":"2505.17082","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-20T12:38:42Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e966723f68ee7a32d8795ba00e46e5c12292b52d25d7137f035f2d572579b79d","abstract_canon_sha256":"e5b2dc78675bd6781f9b5c3abd71191a482aa9575ce62e317285fc39d522316a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:07:46.839360Z","signature_b64":"WPXdVnKR03ayCEW0MbdyXZ43RLziwzBbvFo7tdKx7Q86aGkgjlKVxFg8fow4oEOIv9gGddWDb6IBD4ZG/EggDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5e143d86a455d137fde43873cb7c976afbfd9e3bf27bd743a5877c9abbc3a47c","last_reissued_at":"2026-07-05T11:07:46.838884Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:07:46.838884Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GemMaroc: Unlocking Darija Proficiency in LLMs with Minimal Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Abderrahman Skiredj, Ferdaous Azhari, Houdaifa Atou, Ismail Berrada, Nouamane Tazi","submitted_at":"2025-05-20T12:38:42Z","abstract_excerpt":"Open-source large language models (LLMs) still marginalise Moroccan Arabic (Darija), forcing practitioners either to bolt on heavyweight Arabic adapters or to sacrifice the very reasoning skills that make LLMs useful. We show that a rigorously quality-over-quantity alignment strategy can surface fluent Darija while safeguarding the backbone s cross-lingual reasoning at a sliver of the usual compute. We translate three compact instruction suites LIMA 1 K, DEITA 6 K and TULU 50 K into Darija, preserve 20 of the English originals, and add mathematics, coding and scientific prompts. A LoRA-tuned G"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.17082","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/2505.17082/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":"2505.17082","created_at":"2026-07-05T11:07:46.838943+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.17082v1","created_at":"2026-07-05T11:07:46.838943+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.17082","created_at":"2026-07-05T11:07:46.838943+00:00"},{"alias_kind":"pith_short_12","alias_value":"LYKD3BVEKXIT","created_at":"2026-07-05T11:07:46.838943+00:00"},{"alias_kind":"pith_short_16","alias_value":"LYKD3BVEKXITP7PE","created_at":"2026-07-05T11:07:46.838943+00:00"},{"alias_kind":"pith_short_8","alias_value":"LYKD3BVE","created_at":"2026-07-05T11:07:46.838943+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/LYKD3BVEKXITP7PEHBZ4W7EXNL","json":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL.json","graph_json":"https://pith.science/api/pith-number/LYKD3BVEKXITP7PEHBZ4W7EXNL/graph.json","events_json":"https://pith.science/api/pith-number/LYKD3BVEKXITP7PEHBZ4W7EXNL/events.json","paper":"https://pith.science/paper/LYKD3BVE"},"agent_actions":{"view_html":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL","download_json":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL.json","view_paper":"https://pith.science/paper/LYKD3BVE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.17082&json=true","fetch_graph":"https://pith.science/api/pith-number/LYKD3BVEKXITP7PEHBZ4W7EXNL/graph.json","fetch_events":"https://pith.science/api/pith-number/LYKD3BVEKXITP7PEHBZ4W7EXNL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL/action/storage_attestation","attest_author":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL/action/author_attestation","sign_citation":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL/action/citation_signature","submit_replication":"https://pith.science/pith/LYKD3BVEKXITP7PEHBZ4W7EXNL/action/replication_record"}},"created_at":"2026-07-05T11:07:46.838943+00:00","updated_at":"2026-07-05T11:07:46.838943+00:00"}