{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MKJKNQ6C4ZM4B3O7C46ZBMRB54","short_pith_number":"pith:MKJKNQ6C","schema_version":"1.0","canonical_sha256":"6292a6c3c2e659c0eddf173d90b221ef386065a08afdab6441e29708cfe40cd9","source":{"kind":"arxiv","id":"2504.15120","version":2},"attestation_state":"computed","paper":{"title":"Kuwain 1.5B: An Arabic SLM via Language Injection","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Khalil Hennara, Mohamed Motaism Hamed, Omar Hadid, Safwan AlModhayan, Sara Chrouf, Zeina Aldallal","submitted_at":"2025-04-21T14:17:25Z","abstract_excerpt":"Enhancing existing models with new knowledge is a crucial aspect of AI development. This paper introduces a novel method for integrating a new language into a large language model (LLM). Our approach successfully incorporates a previously unseen target language into an existing LLM without compromising its prior knowledge. We trained a tiny model with 1.5 billion parameters named Kuwain by injecting the Arabic language into a small open-source model mainly trained in English. Our method demonstrates significant improvements in Arabic language performance, with an average 8% improvement 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":"2504.15120","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-21T14:17:25Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1833bc56493fac931065203a918d446fe92c8b9b9820c00534bb0370552bcaf8","abstract_canon_sha256":"50d2afc7e22fbc6f524853f5c93e011db62da6921894835138d1f68d91b1d78f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:56:56.219278Z","signature_b64":"xGSvf9880d33djPbAhWeI3vjvbW4tpnWah1Et5PsovzTqbI3OXVZ+G3XrtVDN8PNvlTmclmSwJnoQBRD0Lk1Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6292a6c3c2e659c0eddf173d90b221ef386065a08afdab6441e29708cfe40cd9","last_reissued_at":"2026-07-05T11:56:56.218795Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:56:56.218795Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Kuwain 1.5B: An Arabic SLM via Language Injection","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Khalil Hennara, Mohamed Motaism Hamed, Omar Hadid, Safwan AlModhayan, Sara Chrouf, Zeina Aldallal","submitted_at":"2025-04-21T14:17:25Z","abstract_excerpt":"Enhancing existing models with new knowledge is a crucial aspect of AI development. This paper introduces a novel method for integrating a new language into a large language model (LLM). Our approach successfully incorporates a previously unseen target language into an existing LLM without compromising its prior knowledge. We trained a tiny model with 1.5 billion parameters named Kuwain by injecting the Arabic language into a small open-source model mainly trained in English. Our method demonstrates significant improvements in Arabic language performance, with an average 8% improvement across "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.15120","kind":"arxiv","version":2},"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/2504.15120/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":"2504.15120","created_at":"2026-07-05T11:56:56.218853+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.15120v2","created_at":"2026-07-05T11:56:56.218853+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.15120","created_at":"2026-07-05T11:56:56.218853+00:00"},{"alias_kind":"pith_short_12","alias_value":"MKJKNQ6C4ZM4","created_at":"2026-07-05T11:56:56.218853+00:00"},{"alias_kind":"pith_short_16","alias_value":"MKJKNQ6C4ZM4B3O7","created_at":"2026-07-05T11:56:56.218853+00:00"},{"alias_kind":"pith_short_8","alias_value":"MKJKNQ6C","created_at":"2026-07-05T11:56:56.218853+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.17894","citing_title":"Mutarjim: Advancing Bidirectional Arabic-English Translation with a Small Language Model","ref_index":10,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54","json":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54.json","graph_json":"https://pith.science/api/pith-number/MKJKNQ6C4ZM4B3O7C46ZBMRB54/graph.json","events_json":"https://pith.science/api/pith-number/MKJKNQ6C4ZM4B3O7C46ZBMRB54/events.json","paper":"https://pith.science/paper/MKJKNQ6C"},"agent_actions":{"view_html":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54","download_json":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54.json","view_paper":"https://pith.science/paper/MKJKNQ6C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.15120&json=true","fetch_graph":"https://pith.science/api/pith-number/MKJKNQ6C4ZM4B3O7C46ZBMRB54/graph.json","fetch_events":"https://pith.science/api/pith-number/MKJKNQ6C4ZM4B3O7C46ZBMRB54/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54/action/storage_attestation","attest_author":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54/action/author_attestation","sign_citation":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54/action/citation_signature","submit_replication":"https://pith.science/pith/MKJKNQ6C4ZM4B3O7C46ZBMRB54/action/replication_record"}},"created_at":"2026-07-05T11:56:56.218853+00:00","updated_at":"2026-07-05T11:56:56.218853+00:00"}