{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:2BAMEOZM4CUUKT6NB7WSM6KYES","short_pith_number":"pith:2BAMEOZM","schema_version":"1.0","canonical_sha256":"d040c23b2ce0a9454fcd0fed26795824b34f3a16d958b9fc4a61e8c519cbad51","source":{"kind":"arxiv","id":"2402.04333","version":3},"attestation_state":"computed","paper":{"title":"LESS: Selecting Influential Data for Targeted Instruction Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Danqi Chen, Mengzhou Xia, Sadhika Malladi, Sanjeev Arora, Suchin Gururangan","submitted_at":"2024-02-06T19:18:04Z","abstract_excerpt":"Instruction tuning has unlocked powerful capabilities in large language models (LLMs), effectively using combined datasets to develop generalpurpose chatbots. However, real-world applications often require a specialized suite of skills (e.g., reasoning). The challenge lies in identifying the most relevant data from these extensive datasets to effectively develop specific capabilities, a setting we frame as targeted instruction tuning. We propose LESS, an optimizer-aware and practically efficient algorithm to effectively estimate data influences and perform Low-rank gradiEnt Similarity Search f"},"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":"2402.04333","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-06T19:18:04Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"7383ec9e54b781b4423cddb2fe5fd0a2c0de584a55b0cb5627f19604c7f3d693","abstract_canon_sha256":"082adf0f5eb8a781fda325bdd087ff058287b86c0812986f7b9533587066381f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:31:11.353491Z","signature_b64":"/sBKPUjCEW7znuaNkcF7HzOV/E/lhkK0zkNoVC+hw3RMiDjGEH0SwXFdb0OP3VL/Vp33vwXFrC/Yb38grzQNAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d040c23b2ce0a9454fcd0fed26795824b34f3a16d958b9fc4a61e8c519cbad51","last_reissued_at":"2026-07-05T08:31:11.352989Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:31:11.352989Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LESS: Selecting Influential Data for Targeted Instruction Tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Danqi Chen, Mengzhou Xia, Sadhika Malladi, Sanjeev Arora, Suchin Gururangan","submitted_at":"2024-02-06T19:18:04Z","abstract_excerpt":"Instruction tuning has unlocked powerful capabilities in large language models (LLMs), effectively using combined datasets to develop generalpurpose chatbots. However, real-world applications often require a specialized suite of skills (e.g., reasoning). The challenge lies in identifying the most relevant data from these extensive datasets to effectively develop specific capabilities, a setting we frame as targeted instruction tuning. We propose LESS, an optimizer-aware and practically efficient algorithm to effectively estimate data influences and perform Low-rank gradiEnt Similarity Search f"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.04333","kind":"arxiv","version":3},"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/2402.04333/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":"2402.04333","created_at":"2026-07-05T08:31:11.353049+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.04333v3","created_at":"2026-07-05T08:31:11.353049+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.04333","created_at":"2026-07-05T08:31:11.353049+00:00"},{"alias_kind":"pith_short_12","alias_value":"2BAMEOZM4CUU","created_at":"2026-07-05T08:31:11.353049+00:00"},{"alias_kind":"pith_short_16","alias_value":"2BAMEOZM4CUUKT6N","created_at":"2026-07-05T08:31:11.353049+00:00"},{"alias_kind":"pith_short_8","alias_value":"2BAMEOZM","created_at":"2026-07-05T08:31:11.353049+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":27,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26091","citing_title":"On-Policy Self-Distillation with Sampled Demonstrations Reduces Output Diversity","ref_index":218,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25178","citing_title":"Transferability for General Reasoning: An Automated Curriculum for Multi-Domain RLVR","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23611","citing_title":"Data Selection Through Iterative Self-Filtering for Vision-Language Settings","ref_index":197,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19549","citing_title":"Predicting Mergeability of Parameter-Efficient Fine-Tuning Updates","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18307","citing_title":"DRIFT: Refining Instruction Data via On-Policy Data Attribution","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2607.02266","citing_title":"HERMES: A Multi-Granularity Labeling Substrate for Pre-training Data Mixtures","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2607.01727","citing_title":"When Does Generating More Help? Disentangling Fixed-Source Synthesis from Source Expansion in Synthetic Data Scaling","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18286","citing_title":"CODEBLOCK: Learning to Supervise Code at the Right Granularity","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23969","citing_title":"SLAP: Stratified Loss-based Pruning for On-Policy Data-Efficient Instruction Tuning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21422","citing_title":"PRISM: Preference-Aware Influence Function Based Data Selection Method for Efficient Fine-Tuning","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25178","citing_title":"Transferability for General Reasoning: An Automated Curriculum for Multi-Domain RLVR","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28631","citing_title":"Single-Rollout Hidden-State Dynamics for Training-Free RLVR Data Selection","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23591","citing_title":"Quantifying the Agreement Between Data-Influence and Data-Similarity to Understand LLM Behavior","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2502.00270","citing_title":"DUET: Optimizing Training Data Mixtures via Feedback from Unseen Evaluation Tasks","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22389","citing_title":"Unified Data Selection for LLM Reasoning","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2508.04149","citing_title":"Difficulty-Based Preference Data Selection by DPO Implicit Reward Gap","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2502.10248","citing_title":"Step-Video-T2V Technical Report: The Practice, Challenges, and Future of Video Foundation Model","ref_index":223,"is_internal_anchor":false},{"citing_arxiv_id":"2508.15202","citing_title":"Fin-PRM: A Domain-Specialized Process Reward Model for Financial Reasoning in Large Language Models","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2601.20375","citing_title":"LLM-AutoDP: Automatic Data Processing via LLM Agents for Model Fine-tuning","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2402.19473","citing_title":"Retrieval-Augmented Generation for AI-Generated Content: A Survey","ref_index":130,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12906","citing_title":"Data Difficulty and the Generalization--Extrapolation Tradeoff in LLM Fine-Tuning","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12681","citing_title":"Toward Communication-Efficient Space Data Centers: Bottlenecks, Architectures, and New Paradigms","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09404","citing_title":"Let the Target Select for Itself: Data Selection via Target-Aligned Paths","ref_index":53,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22374","citing_title":"Selective Contrastive Learning For Gloss Free Sign Language Translation","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05508","citing_title":"Rigorous Interpretation Is a Form of Evaluation","ref_index":45,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES","json":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES.json","graph_json":"https://pith.science/api/pith-number/2BAMEOZM4CUUKT6NB7WSM6KYES/graph.json","events_json":"https://pith.science/api/pith-number/2BAMEOZM4CUUKT6NB7WSM6KYES/events.json","paper":"https://pith.science/paper/2BAMEOZM"},"agent_actions":{"view_html":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES","download_json":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES.json","view_paper":"https://pith.science/paper/2BAMEOZM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.04333&json=true","fetch_graph":"https://pith.science/api/pith-number/2BAMEOZM4CUUKT6NB7WSM6KYES/graph.json","fetch_events":"https://pith.science/api/pith-number/2BAMEOZM4CUUKT6NB7WSM6KYES/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES/action/storage_attestation","attest_author":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES/action/author_attestation","sign_citation":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES/action/citation_signature","submit_replication":"https://pith.science/pith/2BAMEOZM4CUUKT6NB7WSM6KYES/action/replication_record"}},"created_at":"2026-07-05T08:31:11.353049+00:00","updated_at":"2026-07-05T08:31:11.353049+00:00"}