{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:P5WRTV72UJYWFINNUVULPOB4BS","short_pith_number":"pith:P5WRTV72","schema_version":"1.0","canonical_sha256":"7f6d19d7faa27162a1ada568b7b83c0ca9692f6eda2fe5da235fe38ef96d4cff","source":{"kind":"arxiv","id":"2311.00913","version":1},"attestation_state":"computed","paper":{"title":"Self-Influence Guided Data Reweighting for Language Model Pre-training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Megh Thakkar, Partha Talukdar, Sarath Chandar, Shikhar Vashishth, Sriram Ganapathy, Tolga Bolukbasi","submitted_at":"2023-11-02T01:00:46Z","abstract_excerpt":"Language Models (LMs) pre-trained with self-supervision on large text corpora have become the default starting point for developing models for various NLP tasks. Once the pre-training corpus has been assembled, all data samples in the corpus are treated with equal importance during LM pre-training. However, due to varying levels of relevance and quality of data, equal importance to all the data samples may not be the optimal choice. While data reweighting has been explored in the context of task-specific supervised learning and LM fine-tuning, model-driven reweighting for pre-training data has"},"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":"2311.00913","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-11-02T01:00:46Z","cross_cats_sorted":[],"title_canon_sha256":"8c97af78324ecf09ccd40a59b330865651fdf4d8391afb97faef3136f6aeb1f0","abstract_canon_sha256":"80801590315d3ff187237f9ea3ce3e2ffb3dbb3afaee566a9371e4ec211601f2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:08:17.405402Z","signature_b64":"HHTgMiob1ks4veXToWV2wbOhYooI7SnNAICqPg7ErZqFH8WZ0hzwbDUIppet2nn7/hagRtFMeI/GUaqlI0/sBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7f6d19d7faa27162a1ada568b7b83c0ca9692f6eda2fe5da235fe38ef96d4cff","last_reissued_at":"2026-07-05T07:08:17.404909Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:08:17.404909Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Influence Guided Data Reweighting for Language Model Pre-training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Megh Thakkar, Partha Talukdar, Sarath Chandar, Shikhar Vashishth, Sriram Ganapathy, Tolga Bolukbasi","submitted_at":"2023-11-02T01:00:46Z","abstract_excerpt":"Language Models (LMs) pre-trained with self-supervision on large text corpora have become the default starting point for developing models for various NLP tasks. Once the pre-training corpus has been assembled, all data samples in the corpus are treated with equal importance during LM pre-training. However, due to varying levels of relevance and quality of data, equal importance to all the data samples may not be the optimal choice. While data reweighting has been explored in the context of task-specific supervised learning and LM fine-tuning, model-driven reweighting for pre-training data has"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.00913","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/2311.00913/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":"2311.00913","created_at":"2026-07-05T07:08:17.404969+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.00913v1","created_at":"2026-07-05T07:08:17.404969+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.00913","created_at":"2026-07-05T07:08:17.404969+00:00"},{"alias_kind":"pith_short_12","alias_value":"P5WRTV72UJYW","created_at":"2026-07-05T07:08:17.404969+00:00"},{"alias_kind":"pith_short_16","alias_value":"P5WRTV72UJYWFINN","created_at":"2026-07-05T07:08:17.404969+00:00"},{"alias_kind":"pith_short_8","alias_value":"P5WRTV72","created_at":"2026-07-05T07:08:17.404969+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26091","citing_title":"On-Policy Self-Distillation with Sampled Demonstrations Reduces Output Diversity","ref_index":102,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23611","citing_title":"Data Selection Through Iterative Self-Filtering for Vision-Language Settings","ref_index":81,"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":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS","json":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS.json","graph_json":"https://pith.science/api/pith-number/P5WRTV72UJYWFINNUVULPOB4BS/graph.json","events_json":"https://pith.science/api/pith-number/P5WRTV72UJYWFINNUVULPOB4BS/events.json","paper":"https://pith.science/paper/P5WRTV72"},"agent_actions":{"view_html":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS","download_json":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS.json","view_paper":"https://pith.science/paper/P5WRTV72","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.00913&json=true","fetch_graph":"https://pith.science/api/pith-number/P5WRTV72UJYWFINNUVULPOB4BS/graph.json","fetch_events":"https://pith.science/api/pith-number/P5WRTV72UJYWFINNUVULPOB4BS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS/action/storage_attestation","attest_author":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS/action/author_attestation","sign_citation":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS/action/citation_signature","submit_replication":"https://pith.science/pith/P5WRTV72UJYWFINNUVULPOB4BS/action/replication_record"}},"created_at":"2026-07-05T07:08:17.404969+00:00","updated_at":"2026-07-05T07:08:17.404969+00:00"}