{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:CCBKDXAODYBWVL54KBKWAWWFPT","short_pith_number":"pith:CCBKDXAO","schema_version":"1.0","canonical_sha256":"1082a1dc0e1e036aafbc5055605ac57cc396fb0bdc7e45b882de5b2850e71099","source":{"kind":"arxiv","id":"2502.10140","version":1},"attestation_state":"computed","paper":{"title":"Small Models, Big Impact: Efficient Corpus and Graph-Based Adaptation of Small Multilingual Language Models for Low-Resource Languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Daniil Gurgurov, Ivan Vykopal, Josef van Genabith, Simon Ostermann","submitted_at":"2025-02-14T13:10:39Z","abstract_excerpt":"Low-resource languages (LRLs) face significant challenges in natural language processing (NLP) due to limited data. While current state-of-the-art large language models (LLMs) still struggle with LRLs, smaller multilingual models (mLMs) such as mBERT and XLM-R offer greater promise due to a better fit of their capacity to low training data sizes. This study systematically investigates parameter-efficient adapter-based methods for adapting mLMs to LRLs, evaluating three architectures: Sequential Bottleneck, Invertible Bottleneck, and Low-Rank Adaptation. Using unstructured text from GlotCC and "},"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":"2502.10140","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-14T13:10:39Z","cross_cats_sorted":[],"title_canon_sha256":"9131d9a7525e48d1edffc7583e647db58b9653d6aa179e284d633bad81993140","abstract_canon_sha256":"49386e796efc0da7eed93c77be747b30332bf0f27e9981dcb8253d5827508e77"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:14:28.811849Z","signature_b64":"H4kf4UKIWcZHXRp9e5vhfsRgtBBUzRATdz4wtUlxwMqU88LSgUk9/yMhAyMksj5LM+su6vGPQHjoyf5RLbUiBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1082a1dc0e1e036aafbc5055605ac57cc396fb0bdc7e45b882de5b2850e71099","last_reissued_at":"2026-07-05T10:14:28.811274Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:14:28.811274Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Small Models, Big Impact: Efficient Corpus and Graph-Based Adaptation of Small Multilingual Language Models for Low-Resource Languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Daniil Gurgurov, Ivan Vykopal, Josef van Genabith, Simon Ostermann","submitted_at":"2025-02-14T13:10:39Z","abstract_excerpt":"Low-resource languages (LRLs) face significant challenges in natural language processing (NLP) due to limited data. While current state-of-the-art large language models (LLMs) still struggle with LRLs, smaller multilingual models (mLMs) such as mBERT and XLM-R offer greater promise due to a better fit of their capacity to low training data sizes. This study systematically investigates parameter-efficient adapter-based methods for adapting mLMs to LRLs, evaluating three architectures: Sequential Bottleneck, Invertible Bottleneck, and Low-Rank Adaptation. Using unstructured text from GlotCC and "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.10140","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/2502.10140/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":"2502.10140","created_at":"2026-07-05T10:14:28.811364+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.10140v1","created_at":"2026-07-05T10:14:28.811364+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.10140","created_at":"2026-07-05T10:14:28.811364+00:00"},{"alias_kind":"pith_short_12","alias_value":"CCBKDXAODYBW","created_at":"2026-07-05T10:14:28.811364+00:00"},{"alias_kind":"pith_short_16","alias_value":"CCBKDXAODYBWVL54","created_at":"2026-07-05T10:14:28.811364+00:00"},{"alias_kind":"pith_short_8","alias_value":"CCBKDXAO","created_at":"2026-07-05T10:14:28.811364+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18389","citing_title":"Want Better Synthetic Data? Steer It: Activation Steering for Low-Resource Language Generation","ref_index":77,"is_internal_anchor":false},{"citing_arxiv_id":"2408.01119","citing_title":"Task Prompt Vectors: Effective Initialization through Multi-Task Soft-Prompt Transfer","ref_index":16,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT","json":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT.json","graph_json":"https://pith.science/api/pith-number/CCBKDXAODYBWVL54KBKWAWWFPT/graph.json","events_json":"https://pith.science/api/pith-number/CCBKDXAODYBWVL54KBKWAWWFPT/events.json","paper":"https://pith.science/paper/CCBKDXAO"},"agent_actions":{"view_html":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT","download_json":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT.json","view_paper":"https://pith.science/paper/CCBKDXAO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.10140&json=true","fetch_graph":"https://pith.science/api/pith-number/CCBKDXAODYBWVL54KBKWAWWFPT/graph.json","fetch_events":"https://pith.science/api/pith-number/CCBKDXAODYBWVL54KBKWAWWFPT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT/action/storage_attestation","attest_author":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT/action/author_attestation","sign_citation":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT/action/citation_signature","submit_replication":"https://pith.science/pith/CCBKDXAODYBWVL54KBKWAWWFPT/action/replication_record"}},"created_at":"2026-07-05T10:14:28.811364+00:00","updated_at":"2026-07-05T10:14:28.811364+00:00"}