{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:5DN3NABA32BVEJT4ZJ4XYUARA6","short_pith_number":"pith:5DN3NABA","schema_version":"1.0","canonical_sha256":"e8dbb68020de8352267cca797c5011079ebb866d81b5e3a0365f80142fb846e1","source":{"kind":"arxiv","id":"2305.12816","version":1},"attestation_state":"computed","paper":{"title":"Farewell to Aimless Large-scale Pretraining: Influential Subset Selection for Language Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jie Zhou, Junzhe Wang, Menghan Zhang, Qi Zhang, Songyang Gao, Tao Gui, Weikang Zhou, Xiang Gao, Xiao Wang, Yunwen Chen","submitted_at":"2023-05-22T08:18:58Z","abstract_excerpt":"Pretrained language models have achieved remarkable success in various natural language processing tasks. However, pretraining has recently shifted toward larger models and larger data, and this has resulted in significant computational and energy costs. In this paper, we propose Influence Subset Selection (ISS) for language model, which explicitly utilizes end-task knowledge to select a tiny subset of the pretraining corpus. Specifically, the ISS selects the samples that will provide the most positive influence on the performance of the end-task. Furthermore, we design a gradient matching bas"},"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":"2305.12816","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-05-22T08:18:58Z","cross_cats_sorted":[],"title_canon_sha256":"e9cbafa5d1e53000a9f7f9164060dc49e78f3159157f79184183cbddba989c25","abstract_canon_sha256":"19fd6a8874a025501b502632e99ff7c58cd2852a2f4fa7c748b6e53e95fbb322"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:12:15.369085Z","signature_b64":"jHfs0vGKhfrELZIWZjhFvNjdDteoTbgRzGg4eprmXciAqKP8gxpfGObLM0yGwlif1wMQUP2w3KO29cfKKhQKBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e8dbb68020de8352267cca797c5011079ebb866d81b5e3a0365f80142fb846e1","last_reissued_at":"2026-07-05T06:12:15.368621Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:12:15.368621Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Farewell to Aimless Large-scale Pretraining: Influential Subset Selection for Language Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jie Zhou, Junzhe Wang, Menghan Zhang, Qi Zhang, Songyang Gao, Tao Gui, Weikang Zhou, Xiang Gao, Xiao Wang, Yunwen Chen","submitted_at":"2023-05-22T08:18:58Z","abstract_excerpt":"Pretrained language models have achieved remarkable success in various natural language processing tasks. However, pretraining has recently shifted toward larger models and larger data, and this has resulted in significant computational and energy costs. In this paper, we propose Influence Subset Selection (ISS) for language model, which explicitly utilizes end-task knowledge to select a tiny subset of the pretraining corpus. Specifically, the ISS selects the samples that will provide the most positive influence on the performance of the end-task. Furthermore, we design a gradient matching bas"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.12816","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/2305.12816/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":"2305.12816","created_at":"2026-07-05T06:12:15.368681+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.12816v1","created_at":"2026-07-05T06:12:15.368681+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.12816","created_at":"2026-07-05T06:12:15.368681+00:00"},{"alias_kind":"pith_short_12","alias_value":"5DN3NABA32BV","created_at":"2026-07-05T06:12:15.368681+00:00"},{"alias_kind":"pith_short_16","alias_value":"5DN3NABA32BVEJT4","created_at":"2026-07-05T06:12:15.368681+00:00"},{"alias_kind":"pith_short_8","alias_value":"5DN3NABA","created_at":"2026-07-05T06:12:15.368681+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.23969","citing_title":"SLAP: Stratified Loss-based Pruning for On-Policy Data-Efficient Instruction Tuning","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09389","citing_title":"Is More Data Worth the Cost? Dataset Scaling Laws in a Tiny Attention-Only Decoder","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2303.18223","citing_title":"A Survey of Large Language Models","ref_index":255,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6","json":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6.json","graph_json":"https://pith.science/api/pith-number/5DN3NABA32BVEJT4ZJ4XYUARA6/graph.json","events_json":"https://pith.science/api/pith-number/5DN3NABA32BVEJT4ZJ4XYUARA6/events.json","paper":"https://pith.science/paper/5DN3NABA"},"agent_actions":{"view_html":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6","download_json":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6.json","view_paper":"https://pith.science/paper/5DN3NABA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.12816&json=true","fetch_graph":"https://pith.science/api/pith-number/5DN3NABA32BVEJT4ZJ4XYUARA6/graph.json","fetch_events":"https://pith.science/api/pith-number/5DN3NABA32BVEJT4ZJ4XYUARA6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6/action/storage_attestation","attest_author":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6/action/author_attestation","sign_citation":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6/action/citation_signature","submit_replication":"https://pith.science/pith/5DN3NABA32BVEJT4ZJ4XYUARA6/action/replication_record"}},"created_at":"2026-07-05T06:12:15.368681+00:00","updated_at":"2026-07-05T06:12:15.368681+00:00"}