{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MNTIX26JCTCKYDSLE4GUJTPIHO","short_pith_number":"pith:MNTIX26J","schema_version":"1.0","canonical_sha256":"63668bebc914c4ac0e4b270d44cde83b960a9da3e7085957cc266f94fdc2b409","source":{"kind":"arxiv","id":"2410.03115","version":2},"attestation_state":"computed","paper":{"title":"X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scale","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Akiko Eriguchi, Haoran Xu, Hieu Hoang, Huda Khayrallah, Kenton Murray, Philipp Koehn","submitted_at":"2024-10-04T03:17:27Z","abstract_excerpt":"Large language models (LLMs) have achieved remarkable success across various NLP tasks with a focus on English due to English-centric pre-training and limited multilingual data. In this work, we focus on the problem of translation, and while some multilingual LLMs claim to support for hundreds of languages, models often fail to provide high-quality responses for mid- and low-resource languages, leading to imbalanced performance heavily skewed in favor of high-resource languages. We introduce **X-ALMA**, a model designed to ensure top-tier performance across 50 diverse languages, regardless of "},"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":"2410.03115","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-04T03:17:27Z","cross_cats_sorted":[],"title_canon_sha256":"c05389882237e5436948153fb58f684e8cce14864f8945bf5feb46442e158537","abstract_canon_sha256":"5eb78105f2d3c79e8395bd56e2da55cdc1f9e9eb3c4cdd5e1e7f04f33b22ef01"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:22:05.740064Z","signature_b64":"EQJh3AxkPexFHUsygXgpgxIu7MP3cpAKPlQa4JbNI4uLOYSFEpVXEt5ka00l1lxRSmFj1HJ/gzWWUzVT1ZlkDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"63668bebc914c4ac0e4b270d44cde83b960a9da3e7085957cc266f94fdc2b409","last_reissued_at":"2026-07-05T10:22:05.739288Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:22:05.739288Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"X-ALMA: Plug & Play Modules and Adaptive Rejection for Quality Translation at Scale","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Akiko Eriguchi, Haoran Xu, Hieu Hoang, Huda Khayrallah, Kenton Murray, Philipp Koehn","submitted_at":"2024-10-04T03:17:27Z","abstract_excerpt":"Large language models (LLMs) have achieved remarkable success across various NLP tasks with a focus on English due to English-centric pre-training and limited multilingual data. In this work, we focus on the problem of translation, and while some multilingual LLMs claim to support for hundreds of languages, models often fail to provide high-quality responses for mid- and low-resource languages, leading to imbalanced performance heavily skewed in favor of high-resource languages. We introduce **X-ALMA**, a model designed to ensure top-tier performance across 50 diverse languages, regardless of "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.03115","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/2410.03115/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":"2410.03115","created_at":"2026-07-05T10:22:05.739414+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.03115v2","created_at":"2026-07-05T10:22:05.739414+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.03115","created_at":"2026-07-05T10:22:05.739414+00:00"},{"alias_kind":"pith_short_12","alias_value":"MNTIX26JCTCK","created_at":"2026-07-05T10:22:05.739414+00:00"},{"alias_kind":"pith_short_16","alias_value":"MNTIX26JCTCKYDSL","created_at":"2026-07-05T10:22:05.739414+00:00"},{"alias_kind":"pith_short_8","alias_value":"MNTIX26J","created_at":"2026-07-05T10:22:05.739414+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07020","citing_title":"MADE: Beyond Scoring via a Multilingual Agentic Diagnosing Engine for Fine-Grained Evaluation Insights","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05444","citing_title":"Multilingual Coreference Resolution via Cycle-Consistent Machine Translation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17393","citing_title":"Who Watches the Watchmen? Humans Disagree With Translation Metrics on Unseen Domains","ref_index":39,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO","json":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO.json","graph_json":"https://pith.science/api/pith-number/MNTIX26JCTCKYDSLE4GUJTPIHO/graph.json","events_json":"https://pith.science/api/pith-number/MNTIX26JCTCKYDSLE4GUJTPIHO/events.json","paper":"https://pith.science/paper/MNTIX26J"},"agent_actions":{"view_html":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO","download_json":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO.json","view_paper":"https://pith.science/paper/MNTIX26J","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.03115&json=true","fetch_graph":"https://pith.science/api/pith-number/MNTIX26JCTCKYDSLE4GUJTPIHO/graph.json","fetch_events":"https://pith.science/api/pith-number/MNTIX26JCTCKYDSLE4GUJTPIHO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO/action/storage_attestation","attest_author":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO/action/author_attestation","sign_citation":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO/action/citation_signature","submit_replication":"https://pith.science/pith/MNTIX26JCTCKYDSLE4GUJTPIHO/action/replication_record"}},"created_at":"2026-07-05T10:22:05.739414+00:00","updated_at":"2026-07-05T10:22:05.739414+00:00"}