{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:NCTF5MNDTSQX7RVKCBESIC3K4N","short_pith_number":"pith:NCTF5MND","schema_version":"1.0","canonical_sha256":"68a65eb1a39ca17fc6aa1049240b6ae36a2de59417183c31208ca3de85489902","source":{"kind":"arxiv","id":"2309.17425","version":3},"attestation_state":"computed","paper":{"title":"Data Filtering Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Albin Madappally Jose, Alexander Toshev, Alex Fang, Amit Jain, Ludwig Schmidt, Vaishaal Shankar","submitted_at":"2023-09-29T17:37:29Z","abstract_excerpt":"Large training sets have become a cornerstone of machine learning and are the foundation for recent advances in language modeling and multimodal learning. While data curation for pre-training is often still ad-hoc, one common paradigm is to first collect a massive pool of data from the Web and then filter this candidate pool down to an actual training set via various heuristics. In this work, we study the problem of learning a data filtering network (DFN) for this second step of filtering a large uncurated dataset. Our key finding is that the quality of a network for filtering is distinct from"},"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":"2309.17425","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2023-09-29T17:37:29Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"15de1a7cd2396d4249aee61e0598d5906cc4bb5893053ff4ceaea65d0f4ca61b","abstract_canon_sha256":"76ccb0d64fa214d8defb0ba38b359e9893c8250db467673b322d2d2983dfd164"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:09:02.020015Z","signature_b64":"dAm6liGxs5mcauTHITXJNsUSnodCoUGvSsk3kQhAkxRFebO5BjkgqU/MR5doxlMNPlyktKw0KgyYLUxhinffBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"68a65eb1a39ca17fc6aa1049240b6ae36a2de59417183c31208ca3de85489902","last_reissued_at":"2026-07-05T07:09:02.019524Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:09:02.019524Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Data Filtering Networks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Albin Madappally Jose, Alexander Toshev, Alex Fang, Amit Jain, Ludwig Schmidt, Vaishaal Shankar","submitted_at":"2023-09-29T17:37:29Z","abstract_excerpt":"Large training sets have become a cornerstone of machine learning and are the foundation for recent advances in language modeling and multimodal learning. While data curation for pre-training is often still ad-hoc, one common paradigm is to first collect a massive pool of data from the Web and then filter this candidate pool down to an actual training set via various heuristics. In this work, we study the problem of learning a data filtering network (DFN) for this second step of filtering a large uncurated dataset. Our key finding is that the quality of a network for filtering is distinct from"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2309.17425","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/2309.17425/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":"2309.17425","created_at":"2026-07-05T07:09:02.019580+00:00"},{"alias_kind":"arxiv_version","alias_value":"2309.17425v3","created_at":"2026-07-05T07:09:02.019580+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2309.17425","created_at":"2026-07-05T07:09:02.019580+00:00"},{"alias_kind":"pith_short_12","alias_value":"NCTF5MNDTSQX","created_at":"2026-07-05T07:09:02.019580+00:00"},{"alias_kind":"pith_short_16","alias_value":"NCTF5MNDTSQX7RVK","created_at":"2026-07-05T07:09:02.019580+00:00"},{"alias_kind":"pith_short_8","alias_value":"NCTF5MND","created_at":"2026-07-05T07:09:02.019580+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":34,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26091","citing_title":"On-Policy Self-Distillation with Sampled Demonstrations Reduces Output Diversity","ref_index":221,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26199","citing_title":"MIRAGE: Protecting against Malicious Image Editing via False Moderation","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26794","citing_title":"ReasonCLIP-58M: Visually Grounded Commonsense Reasoning Supervision for CLIP","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27313","citing_title":"ViQ: Text-Aligned Visual Quantized Representations at Any Resolution","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23611","citing_title":"Data Selection Through Iterative Self-Filtering for Vision-Language Settings","ref_index":200,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21734","citing_title":"HPP: Hierarchical Programmatic Probing for Long Video Understanding by Decoupling Perception and Reasoning","ref_index":239,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19584","citing_title":"Language-Instructed Vision Embeddings for Controllable and Generalizable Perception","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17030","citing_title":"Qwen-RobotWorld Technical Report: Unifying Embodied World Modeling through Language-Conditioned Video Generation","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13289","citing_title":"HYDRA-X: Native Unified Multimodal Models with Holistic Visual Tokenizers","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28551","citing_title":"DataComp-VLM: Improved Open Datasets for Vision-Language Models","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00174","citing_title":"Steal the Patch Size: Adversarially Manipulate Vision-Language Models","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07032","citing_title":"Never Seen Before: Benchmarking Genuine Zero-Shot Composed Image Retrieval with Consistent Video-Sourced Datasets","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00684","citing_title":"AdaBoosting Text Prompts for Vision-Language Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28551","citing_title":"DataComp-VLM: Improved Open Datasets for Vision-Language Models","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27313","citing_title":"ViQ: Text-Aligned Visual Quantized Representations at Any Resolution","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05204","citing_title":"D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26199","citing_title":"MIRAGE: Protecting against Malicious Image Editing via False Moderation","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22651","citing_title":"What Does the Caption Really Say? Counterfactual Phrase Intervention for Compositional Data Selection in Vision-Language Pretraining","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05204","citing_title":"D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2406.11794","citing_title":"DataComp-LM: In search of the next generation of training sets for language models","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2512.20157","citing_title":"SigLino: Efficient Multi-Teacher Distillation for Agglomerative Vision Foundation Models","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2403.09611","citing_title":"MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02546","citing_title":"RGB-Pointmap Pretraining for Unified 3D Scene Understanding","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12491","citing_title":"Elastic Attention Cores for Scalable Vision Transformers","ref_index":132,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24997","citing_title":"DouC: Dual-Branch CLIP for Training-Free Open-Vocabulary Segmentation","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N","json":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N.json","graph_json":"https://pith.science/api/pith-number/NCTF5MNDTSQX7RVKCBESIC3K4N/graph.json","events_json":"https://pith.science/api/pith-number/NCTF5MNDTSQX7RVKCBESIC3K4N/events.json","paper":"https://pith.science/paper/NCTF5MND"},"agent_actions":{"view_html":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N","download_json":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N.json","view_paper":"https://pith.science/paper/NCTF5MND","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2309.17425&json=true","fetch_graph":"https://pith.science/api/pith-number/NCTF5MNDTSQX7RVKCBESIC3K4N/graph.json","fetch_events":"https://pith.science/api/pith-number/NCTF5MNDTSQX7RVKCBESIC3K4N/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N/action/storage_attestation","attest_author":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N/action/author_attestation","sign_citation":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N/action/citation_signature","submit_replication":"https://pith.science/pith/NCTF5MNDTSQX7RVKCBESIC3K4N/action/replication_record"}},"created_at":"2026-07-05T07:09:02.019580+00:00","updated_at":"2026-07-05T07:09:02.019580+00:00"}