{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:GF2WXYTCQF7EDREO2ENNR6D3EU","short_pith_number":"pith:GF2WXYTC","schema_version":"1.0","canonical_sha256":"31756be262817e41c48ed11ad8f87b2524e9afa5eb76aed715376c75a4337723","source":{"kind":"arxiv","id":"2402.10430","version":1},"attestation_state":"computed","paper":{"title":"Smaller Language Models are capable of selecting Instruction-Tuning Training Data for Larger Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alex Nguyen, Dheeraj Mekala, Jingbo Shang","submitted_at":"2024-02-16T03:39:37Z","abstract_excerpt":"Instruction-tuning language models has become a crucial step in aligning them for general use. Typically, this process involves extensive training on large datasets, incurring high training costs. In this paper, we introduce a novel training data selection based on the learning percentage of the samples. We assert that current language models possess the capability to autonomously select high-quality training data, leading to comparable or improved performance compared to training on the entire dataset. Our experiments span different-sized models, revealing that this characteristic holds for m"},"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":"2402.10430","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-16T03:39:37Z","cross_cats_sorted":[],"title_canon_sha256":"5b778c6a27738bc0ef18bd25f8fe2084d2a393f893ffe4bb1c949b7f90955fda","abstract_canon_sha256":"524bceac18820ba9c8ebef90d55f53639bb32e08166b7a51c40bf6c6ece81cc1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:45:56.299443Z","signature_b64":"Q2YQ2kdJxEp6c38uxxxE80ZUL4kKZNTh4I9X2M350JpXQnSy05bkBpqPJ8wb2/8DwwUjnZIcOTaUXQ5YMAB8BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"31756be262817e41c48ed11ad8f87b2524e9afa5eb76aed715376c75a4337723","last_reissued_at":"2026-07-05T07:45:56.298894Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:45:56.298894Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Smaller Language Models are capable of selecting Instruction-Tuning Training Data for Larger Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alex Nguyen, Dheeraj Mekala, Jingbo Shang","submitted_at":"2024-02-16T03:39:37Z","abstract_excerpt":"Instruction-tuning language models has become a crucial step in aligning them for general use. Typically, this process involves extensive training on large datasets, incurring high training costs. In this paper, we introduce a novel training data selection based on the learning percentage of the samples. We assert that current language models possess the capability to autonomously select high-quality training data, leading to comparable or improved performance compared to training on the entire dataset. Our experiments span different-sized models, revealing that this characteristic holds for m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.10430","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/2402.10430/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":"2402.10430","created_at":"2026-07-05T07:45:56.298963+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.10430v1","created_at":"2026-07-05T07:45:56.298963+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.10430","created_at":"2026-07-05T07:45:56.298963+00:00"},{"alias_kind":"pith_short_12","alias_value":"GF2WXYTCQF7E","created_at":"2026-07-05T07:45:56.298963+00:00"},{"alias_kind":"pith_short_16","alias_value":"GF2WXYTCQF7EDREO","created_at":"2026-07-05T07:45:56.298963+00:00"},{"alias_kind":"pith_short_8","alias_value":"GF2WXYTC","created_at":"2026-07-05T07:45:56.298963+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.11810","citing_title":"GRACE: A Dynamic Coreset Selection Framework for Large Language Model Optimization","ref_index":58,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU","json":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU.json","graph_json":"https://pith.science/api/pith-number/GF2WXYTCQF7EDREO2ENNR6D3EU/graph.json","events_json":"https://pith.science/api/pith-number/GF2WXYTCQF7EDREO2ENNR6D3EU/events.json","paper":"https://pith.science/paper/GF2WXYTC"},"agent_actions":{"view_html":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU","download_json":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU.json","view_paper":"https://pith.science/paper/GF2WXYTC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.10430&json=true","fetch_graph":"https://pith.science/api/pith-number/GF2WXYTCQF7EDREO2ENNR6D3EU/graph.json","fetch_events":"https://pith.science/api/pith-number/GF2WXYTCQF7EDREO2ENNR6D3EU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU/action/storage_attestation","attest_author":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU/action/author_attestation","sign_citation":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU/action/citation_signature","submit_replication":"https://pith.science/pith/GF2WXYTCQF7EDREO2ENNR6D3EU/action/replication_record"}},"created_at":"2026-07-05T07:45:56.298963+00:00","updated_at":"2026-07-05T07:45:56.298963+00:00"}