{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:NIPY4YKV6VG22N5FPHPBF25FGI","short_pith_number":"pith:NIPY4YKV","schema_version":"1.0","canonical_sha256":"6a1f8e6155f54dad37a579de12eba5322fe6aa25c71ed25facc8cc83fccce8db","source":{"kind":"arxiv","id":"2503.01807","version":2},"attestation_state":"computed","paper":{"title":"Large-Scale Data Selection for Instruction Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Faeze Brahman, Hamish Ivison, Muru Zhang, Pang Wei Koh, Pradeep Dasigi","submitted_at":"2025-03-03T18:37:26Z","abstract_excerpt":"Selecting high-quality training data from a larger pool is a crucial step when instruction-tuning language models, as carefully curated datasets often produce models that outperform those trained on much larger, noisier datasets. Automated data selection approaches for instruction-tuning are typically tested by selecting small datasets (roughly 10k samples) from small pools (100-200k samples). However, popular deployed instruction-tuned models often train on hundreds of thousands to millions of samples, subsampled from even larger data pools. We present a systematic study of how well data sele"},"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":"2503.01807","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-03T18:37:26Z","cross_cats_sorted":[],"title_canon_sha256":"d47c7fc6b2865b33b4a7473267e810d77e47ffefe3e48eec77527fe6edbeb8c7","abstract_canon_sha256":"bdf3f74ac8bb03f8ec6fde5cf08680b00fdafdf206b1d8dc7723d906e20e0e20"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:24:15.664078Z","signature_b64":"soeYA+rD79G7W0rBOubXpjnSsjWKEytq4IC6Tcm9pqY7JyF7JE+9LRpi9eoTa2MzNiIZzWCWNs//InBmysg9Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6a1f8e6155f54dad37a579de12eba5322fe6aa25c71ed25facc8cc83fccce8db","last_reissued_at":"2026-07-05T11:24:15.663523Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:24:15.663523Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large-Scale Data Selection for Instruction Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Faeze Brahman, Hamish Ivison, Muru Zhang, Pang Wei Koh, Pradeep Dasigi","submitted_at":"2025-03-03T18:37:26Z","abstract_excerpt":"Selecting high-quality training data from a larger pool is a crucial step when instruction-tuning language models, as carefully curated datasets often produce models that outperform those trained on much larger, noisier datasets. Automated data selection approaches for instruction-tuning are typically tested by selecting small datasets (roughly 10k samples) from small pools (100-200k samples). However, popular deployed instruction-tuned models often train on hundreds of thousands to millions of samples, subsampled from even larger data pools. We present a systematic study of how well data sele"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.01807","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/2503.01807/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":"2503.01807","created_at":"2026-07-05T11:24:15.663586+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.01807v2","created_at":"2026-07-05T11:24:15.663586+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.01807","created_at":"2026-07-05T11:24:15.663586+00:00"},{"alias_kind":"pith_short_12","alias_value":"NIPY4YKV6VG2","created_at":"2026-07-05T11:24:15.663586+00:00"},{"alias_kind":"pith_short_16","alias_value":"NIPY4YKV6VG22N5F","created_at":"2026-07-05T11:24:15.663586+00:00"},{"alias_kind":"pith_short_8","alias_value":"NIPY4YKV","created_at":"2026-07-05T11:24:15.663586+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18307","citing_title":"DRIFT: Refining Instruction Data via On-Policy Data Attribution","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13668","citing_title":"Influcoder: Distilling Decoders' Gradient Influence Rankings into an Encoder for Data Attribution","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2602.18584","citing_title":"GIST: Targeted Data Selection for Instruction Tuning via Coupled Optimization Geometry","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15675","citing_title":"Interaction-Aware Influence Functions for Group Attribution","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2602.03452","citing_title":"Beyond Variance: Prompt-Efficient RLVR via Rare-Event Amplification and Bidirectional Pairing","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2504.20571","citing_title":"Reinforcement Learning for Reasoning in Large Language Models with One Training Example","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12944","citing_title":"From Instance Selection to Fixed-Pool Data Recipe Search for Supervised Fine-Tuning","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05227","citing_title":"Rethinking Data Curation in LLM Training: Online Reweighting Offers Better Generalization than Offline Methods","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17928","citing_title":"HEALing Entropy Collapse: Enhancing Exploration in Few-Shot RLVR via Hybrid-Domain Entropy Dynamics Alignment","ref_index":92,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI","json":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI.json","graph_json":"https://pith.science/api/pith-number/NIPY4YKV6VG22N5FPHPBF25FGI/graph.json","events_json":"https://pith.science/api/pith-number/NIPY4YKV6VG22N5FPHPBF25FGI/events.json","paper":"https://pith.science/paper/NIPY4YKV"},"agent_actions":{"view_html":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI","download_json":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI.json","view_paper":"https://pith.science/paper/NIPY4YKV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.01807&json=true","fetch_graph":"https://pith.science/api/pith-number/NIPY4YKV6VG22N5FPHPBF25FGI/graph.json","fetch_events":"https://pith.science/api/pith-number/NIPY4YKV6VG22N5FPHPBF25FGI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI/action/storage_attestation","attest_author":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI/action/author_attestation","sign_citation":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI/action/citation_signature","submit_replication":"https://pith.science/pith/NIPY4YKV6VG22N5FPHPBF25FGI/action/replication_record"}},"created_at":"2026-07-05T11:24:15.663586+00:00","updated_at":"2026-07-05T11:24:15.663586+00:00"}