{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FNOSBMX2JYMX6UO2M22TICQYTQ","short_pith_number":"pith:FNOSBMX2","schema_version":"1.0","canonical_sha256":"2b5d20b2fa4e197f51da66b5340a189c3d8238aac93acefac948c4c377f7888d","source":{"kind":"arxiv","id":"2506.17080","version":1},"attestation_state":"computed","paper":{"title":"Tower+: Bridging Generality and Translation Specialization in Multilingual LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Amin Farajian, Andr\\'e F. T. Martins, Jo\\~ao Alves, Jos\\'e Pombal, Nuno M. Guerreiro, Pedro Teixeirinha, Ricardo Rei","submitted_at":"2025-06-20T15:30:06Z","abstract_excerpt":"Fine-tuning pretrained LLMs has been shown to be an effective strategy for reaching state-of-the-art performance on specific tasks like machine translation. However, this process of adaptation often implies sacrificing general-purpose capabilities, such as conversational reasoning and instruction-following, hampering the utility of the system in real-world applications that require a mixture of skills. In this paper, we introduce Tower+, a suite of models designed to deliver strong performance across both translation and multilingual general-purpose text capabilities. We achieve a Pareto front"},"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":"2506.17080","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-20T15:30:06Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9bdc7ef498357a932e300ba19b13686cd163f4aaa4250527bcf14d33e26ac2d9","abstract_canon_sha256":"0d6baa96ce680777aa1d8fca18b60b253bab4497b54af71cd82731c575ed525a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:24:56.116583Z","signature_b64":"6uXppw4mUyd/P6QXjGEJHx+24qTGNbAE9KbqJ6tdrQ8j5phWVSrtjnjswGROEA4rdgEmu95/IEzWoEjbB9UlDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2b5d20b2fa4e197f51da66b5340a189c3d8238aac93acefac948c4c377f7888d","last_reissued_at":"2026-07-05T11:24:56.116095Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:24:56.116095Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tower+: Bridging Generality and Translation Specialization in Multilingual LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Amin Farajian, Andr\\'e F. T. Martins, Jo\\~ao Alves, Jos\\'e Pombal, Nuno M. Guerreiro, Pedro Teixeirinha, Ricardo Rei","submitted_at":"2025-06-20T15:30:06Z","abstract_excerpt":"Fine-tuning pretrained LLMs has been shown to be an effective strategy for reaching state-of-the-art performance on specific tasks like machine translation. However, this process of adaptation often implies sacrificing general-purpose capabilities, such as conversational reasoning and instruction-following, hampering the utility of the system in real-world applications that require a mixture of skills. In this paper, we introduce Tower+, a suite of models designed to deliver strong performance across both translation and multilingual general-purpose text capabilities. We achieve a Pareto front"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.17080","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/2506.17080/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":"2506.17080","created_at":"2026-07-05T11:24:56.116155+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.17080v1","created_at":"2026-07-05T11:24:56.116155+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.17080","created_at":"2026-07-05T11:24:56.116155+00:00"},{"alias_kind":"pith_short_12","alias_value":"FNOSBMX2JYMX","created_at":"2026-07-05T11:24:56.116155+00:00"},{"alias_kind":"pith_short_16","alias_value":"FNOSBMX2JYMX6UO2","created_at":"2026-07-05T11:24:56.116155+00:00"},{"alias_kind":"pith_short_8","alias_value":"FNOSBMX2","created_at":"2026-07-05T11:24:56.116155+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06177","citing_title":"Ouvia: A User-centered Framework for Measuring Usability of Speech Translation in Real-World Communication Scenarios","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17255","citing_title":"MLLP-VRAIN UPV system for the IWSLT 2026 Simultaneous Speech Translation task","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2504.01919","citing_title":"Bridging the Linguistic Divide: A Survey on Leveraging Large Language Models for Machine Translation","ref_index":104,"is_internal_anchor":false},{"citing_arxiv_id":"2512.16378","citing_title":"Hearing to Translate: The Effectiveness of Speech Modality Integration into LLMs","ref_index":77,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ","json":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ.json","graph_json":"https://pith.science/api/pith-number/FNOSBMX2JYMX6UO2M22TICQYTQ/graph.json","events_json":"https://pith.science/api/pith-number/FNOSBMX2JYMX6UO2M22TICQYTQ/events.json","paper":"https://pith.science/paper/FNOSBMX2"},"agent_actions":{"view_html":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ","download_json":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ.json","view_paper":"https://pith.science/paper/FNOSBMX2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.17080&json=true","fetch_graph":"https://pith.science/api/pith-number/FNOSBMX2JYMX6UO2M22TICQYTQ/graph.json","fetch_events":"https://pith.science/api/pith-number/FNOSBMX2JYMX6UO2M22TICQYTQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ/action/storage_attestation","attest_author":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ/action/author_attestation","sign_citation":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ/action/citation_signature","submit_replication":"https://pith.science/pith/FNOSBMX2JYMX6UO2M22TICQYTQ/action/replication_record"}},"created_at":"2026-07-05T11:24:56.116155+00:00","updated_at":"2026-07-05T11:24:56.116155+00:00"}