{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ZCYPOSVLFNEO4E47MJP7SPF2Z3","short_pith_number":"pith:ZCYPOSVL","schema_version":"1.0","canonical_sha256":"c8b0f74aab2b48ee139f625ff93cbacec84038333b5a2cb2d017eaa614b2e5cf","source":{"kind":"arxiv","id":"2309.08708","version":2},"attestation_state":"computed","paper":{"title":"Vocabulary-level Memory Efficiency for Language Model Fine-tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Miles Williams, Nikolaos Aletras","submitted_at":"2023-09-15T19:00:00Z","abstract_excerpt":"The extensive memory footprint of language model (LM) fine-tuning poses a challenge for both researchers and practitioners. LMs use an embedding matrix to represent extensive vocabularies, forming a substantial proportion of the model parameters. While previous work towards memory-efficient fine-tuning has focused on minimizing the number of trainable parameters, reducing the memory footprint of the embedding matrix has yet to be explored. We first demonstrate that a significant proportion of the vocabulary remains unused during fine-tuning. We then propose a simple yet effective approach that"},"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":"2309.08708","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-09-15T19:00:00Z","cross_cats_sorted":[],"title_canon_sha256":"e75f32beed959d490c0da83975edd10103e76d15fb79160e0f8e0616871a8843","abstract_canon_sha256":"d824031d85f3dfeac3db1572ab5f2be976a5b12da0b3a783fc4cd1bf36049ef8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:38:31.106759Z","signature_b64":"yrGgoyVtU1fxmNkh7pVvtv7Oy/yVygGjsvzNKhJF6wkqduDGtZY2Jo50Sgg3KvPOgXyltkKYY9TzaIMw1KtmDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c8b0f74aab2b48ee139f625ff93cbacec84038333b5a2cb2d017eaa614b2e5cf","last_reissued_at":"2026-07-05T10:38:31.106347Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:38:31.106347Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Vocabulary-level Memory Efficiency for Language Model Fine-tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Miles Williams, Nikolaos Aletras","submitted_at":"2023-09-15T19:00:00Z","abstract_excerpt":"The extensive memory footprint of language model (LM) fine-tuning poses a challenge for both researchers and practitioners. LMs use an embedding matrix to represent extensive vocabularies, forming a substantial proportion of the model parameters. While previous work towards memory-efficient fine-tuning has focused on minimizing the number of trainable parameters, reducing the memory footprint of the embedding matrix has yet to be explored. We first demonstrate that a significant proportion of the vocabulary remains unused during fine-tuning. We then propose a simple yet effective approach that"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2309.08708","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/2309.08708/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":"2309.08708","created_at":"2026-07-05T10:38:31.106407+00:00"},{"alias_kind":"arxiv_version","alias_value":"2309.08708v2","created_at":"2026-07-05T10:38:31.106407+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2309.08708","created_at":"2026-07-05T10:38:31.106407+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZCYPOSVLFNEO","created_at":"2026-07-05T10:38:31.106407+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZCYPOSVLFNEO4E47","created_at":"2026-07-05T10:38:31.106407+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZCYPOSVL","created_at":"2026-07-05T10:38:31.106407+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.16323","citing_title":"SpeLLM: Character-Level Multi-Head Decoding","ref_index":23,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3","json":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3.json","graph_json":"https://pith.science/api/pith-number/ZCYPOSVLFNEO4E47MJP7SPF2Z3/graph.json","events_json":"https://pith.science/api/pith-number/ZCYPOSVLFNEO4E47MJP7SPF2Z3/events.json","paper":"https://pith.science/paper/ZCYPOSVL"},"agent_actions":{"view_html":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3","download_json":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3.json","view_paper":"https://pith.science/paper/ZCYPOSVL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2309.08708&json=true","fetch_graph":"https://pith.science/api/pith-number/ZCYPOSVLFNEO4E47MJP7SPF2Z3/graph.json","fetch_events":"https://pith.science/api/pith-number/ZCYPOSVLFNEO4E47MJP7SPF2Z3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3/action/storage_attestation","attest_author":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3/action/author_attestation","sign_citation":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3/action/citation_signature","submit_replication":"https://pith.science/pith/ZCYPOSVLFNEO4E47MJP7SPF2Z3/action/replication_record"}},"created_at":"2026-07-05T10:38:31.106407+00:00","updated_at":"2026-07-05T10:38:31.106407+00:00"}