{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:X6PZUWQKGP4UTCEKFCOLIRRUZT","short_pith_number":"pith:X6PZUWQK","schema_version":"1.0","canonical_sha256":"bf9f9a5a0a33f949888a289cb44634ccf08a80a4634a9d4ae57a27dcc21f8e10","source":{"kind":"arxiv","id":"2502.18020","version":1},"attestation_state":"computed","paper":{"title":"AfroXLMR-Comet: Multilingual Knowledge Distillation with Attention Matching for Low-Resource languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Jaskaran Singh Walia, Joshua Sakthivel Raju, Sanjay S, Srinivas Raghav, Vukosi Marivate","submitted_at":"2025-02-25T09:28:47Z","abstract_excerpt":"Language model compression through knowledge distillation has emerged as a promising approach for deploying large language models in resource-constrained environments. However, existing methods often struggle to maintain performance when distilling multilingual models, especially for low-resource languages. In this paper, we present a novel hybrid distillation approach that combines traditional knowledge distillation with a simplified attention matching mechanism, specifically designed for multilingual contexts. Our method introduces an extremely compact student model architecture, significant"},"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":"2502.18020","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-25T09:28:47Z","cross_cats_sorted":["cs.AI","cs.IR","cs.LG"],"title_canon_sha256":"90406067e99b49d0f10dad846be62123198cf19715d1e23027addbee1611a370","abstract_canon_sha256":"43b7f0f802ec3afa7e560b2dde1a8002f1531a9cf84dda0c39dc8143f8f18699"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:19:47.814115Z","signature_b64":"qN+sI46nPXmFCYwF9odPU7VbYQ7gcrhU6yFpkRb/oMYV9qsc0Y7PcG+Q0eK0hIEbEaHWG8ML2Hu7RFck2hqfCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bf9f9a5a0a33f949888a289cb44634ccf08a80a4634a9d4ae57a27dcc21f8e10","last_reissued_at":"2026-07-05T10:19:47.813634Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:19:47.813634Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AfroXLMR-Comet: Multilingual Knowledge Distillation with Attention Matching for Low-Resource languages","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Jaskaran Singh Walia, Joshua Sakthivel Raju, Sanjay S, Srinivas Raghav, Vukosi Marivate","submitted_at":"2025-02-25T09:28:47Z","abstract_excerpt":"Language model compression through knowledge distillation has emerged as a promising approach for deploying large language models in resource-constrained environments. However, existing methods often struggle to maintain performance when distilling multilingual models, especially for low-resource languages. In this paper, we present a novel hybrid distillation approach that combines traditional knowledge distillation with a simplified attention matching mechanism, specifically designed for multilingual contexts. Our method introduces an extremely compact student model architecture, significant"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.18020","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/2502.18020/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":"2502.18020","created_at":"2026-07-05T10:19:47.813692+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.18020v1","created_at":"2026-07-05T10:19:47.813692+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.18020","created_at":"2026-07-05T10:19:47.813692+00:00"},{"alias_kind":"pith_short_12","alias_value":"X6PZUWQKGP4U","created_at":"2026-07-05T10:19:47.813692+00:00"},{"alias_kind":"pith_short_16","alias_value":"X6PZUWQKGP4UTCEK","created_at":"2026-07-05T10:19:47.813692+00:00"},{"alias_kind":"pith_short_8","alias_value":"X6PZUWQK","created_at":"2026-07-05T10:19:47.813692+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.10920","citing_title":"HanjaBridge: Resolving Semantic Ambiguity in Korean LLMs via Hanja-Augmented Pre-Training","ref_index":25,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT","json":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT.json","graph_json":"https://pith.science/api/pith-number/X6PZUWQKGP4UTCEKFCOLIRRUZT/graph.json","events_json":"https://pith.science/api/pith-number/X6PZUWQKGP4UTCEKFCOLIRRUZT/events.json","paper":"https://pith.science/paper/X6PZUWQK"},"agent_actions":{"view_html":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT","download_json":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT.json","view_paper":"https://pith.science/paper/X6PZUWQK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.18020&json=true","fetch_graph":"https://pith.science/api/pith-number/X6PZUWQKGP4UTCEKFCOLIRRUZT/graph.json","fetch_events":"https://pith.science/api/pith-number/X6PZUWQKGP4UTCEKFCOLIRRUZT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT/action/storage_attestation","attest_author":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT/action/author_attestation","sign_citation":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT/action/citation_signature","submit_replication":"https://pith.science/pith/X6PZUWQKGP4UTCEKFCOLIRRUZT/action/replication_record"}},"created_at":"2026-07-05T10:19:47.813692+00:00","updated_at":"2026-07-05T10:19:47.813692+00:00"}