{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BMAWQATVVMA4ZVGGBPBMBFFNDG","short_pith_number":"pith:BMAWQATV","schema_version":"1.0","canonical_sha256":"0b01680275ab01ccd4c60bc2c094ad19b06c50d5722c3efe34bbbf6631b19b32","source":{"kind":"arxiv","id":"2505.21171","version":1},"attestation_state":"computed","paper":{"title":"M-Wanda: Improving One-Shot Pruning for Multilingual LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ivan Titov, Rochelle Choenni","submitted_at":"2025-05-27T13:24:38Z","abstract_excerpt":"Multilingual LLM performance is often critically dependent on model size. With an eye on efficiency, this has led to a surge in interest in one-shot pruning methods that retain the benefits of large-scale pretraining while shrinking the model size. However, as pruning tends to come with performance loss, it is important to understand the trade-offs between multilinguality and sparsification. In this work, we study multilingual performance under different sparsity constraints and show that moderate ratios already substantially harm performance. To help bridge this gap, we propose M-Wanda, a pru"},"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":"2505.21171","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-05-27T13:24:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"bfce24a23530194a8e16a3b24e2f87d89e6d65deafc706b721678ede6a6e4e0f","abstract_canon_sha256":"05c2cc3a90c1e79c5468654ef4cd7a912d2fd974771b51db8b7d3bfaad98f19e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:10:24.903193Z","signature_b64":"592qjZaI+6v1dgO5ycVBsGjhrt31hPCjzmLzU4w+iRrqPDCYtP1TASp71wkmbpl+bUe25dw2N5YKoms/IJzwDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0b01680275ab01ccd4c60bc2c094ad19b06c50d5722c3efe34bbbf6631b19b32","last_reissued_at":"2026-07-05T11:10:24.902779Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:10:24.902779Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"M-Wanda: Improving One-Shot Pruning for Multilingual LLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ivan Titov, Rochelle Choenni","submitted_at":"2025-05-27T13:24:38Z","abstract_excerpt":"Multilingual LLM performance is often critically dependent on model size. With an eye on efficiency, this has led to a surge in interest in one-shot pruning methods that retain the benefits of large-scale pretraining while shrinking the model size. However, as pruning tends to come with performance loss, it is important to understand the trade-offs between multilinguality and sparsification. In this work, we study multilingual performance under different sparsity constraints and show that moderate ratios already substantially harm performance. To help bridge this gap, we propose M-Wanda, a pru"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.21171","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/2505.21171/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":"2505.21171","created_at":"2026-07-05T11:10:24.902835+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.21171v1","created_at":"2026-07-05T11:10:24.902835+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.21171","created_at":"2026-07-05T11:10:24.902835+00:00"},{"alias_kind":"pith_short_12","alias_value":"BMAWQATVVMA4","created_at":"2026-07-05T11:10:24.902835+00:00"},{"alias_kind":"pith_short_16","alias_value":"BMAWQATVVMA4ZVGG","created_at":"2026-07-05T11:10:24.902835+00:00"},{"alias_kind":"pith_short_8","alias_value":"BMAWQATV","created_at":"2026-07-05T11:10:24.902835+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG","json":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG.json","graph_json":"https://pith.science/api/pith-number/BMAWQATVVMA4ZVGGBPBMBFFNDG/graph.json","events_json":"https://pith.science/api/pith-number/BMAWQATVVMA4ZVGGBPBMBFFNDG/events.json","paper":"https://pith.science/paper/BMAWQATV"},"agent_actions":{"view_html":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG","download_json":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG.json","view_paper":"https://pith.science/paper/BMAWQATV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.21171&json=true","fetch_graph":"https://pith.science/api/pith-number/BMAWQATVVMA4ZVGGBPBMBFFNDG/graph.json","fetch_events":"https://pith.science/api/pith-number/BMAWQATVVMA4ZVGGBPBMBFFNDG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG/action/storage_attestation","attest_author":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG/action/author_attestation","sign_citation":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG/action/citation_signature","submit_replication":"https://pith.science/pith/BMAWQATVVMA4ZVGGBPBMBFFNDG/action/replication_record"}},"created_at":"2026-07-05T11:10:24.902835+00:00","updated_at":"2026-07-05T11:10:24.902835+00:00"}