{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:S66X43R6VHVRXX73QREKVXHVJR","short_pith_number":"pith:S66X43R6","schema_version":"1.0","canonical_sha256":"97bd7e6e3ea9eb1bdffb8448aadcf54c53ef243c54d2772c68f9d00375a86745","source":{"kind":"arxiv","id":"2505.06356","version":1},"attestation_state":"computed","paper":{"title":"Understanding and Mitigating Toxicity in Image-Text Pretraining Datasets: A Case Study on LLaVA","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Karthik Reddy Kanjula, Nahid Alam, Shayekh Bin Islam, Surya Guthikonda","submitted_at":"2025-05-09T18:01:50Z","abstract_excerpt":"Pretraining datasets are foundational to the development of multimodal models, yet they often have inherent biases and toxic content from the web-scale corpora they are sourced from. In this paper, we investigate the prevalence of toxicity in LLaVA image-text pretraining dataset, examining how harmful content manifests in different modalities. We present a comprehensive analysis of common toxicity categories and propose targeted mitigation strategies, resulting in the creation of a refined toxicity-mitigated dataset. This dataset removes 7,531 of toxic image-text pairs in the LLaVA pre-trainin"},"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":"2505.06356","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-09T18:01:50Z","cross_cats_sorted":[],"title_canon_sha256":"e861196fafcfdd9dd713afc34995b279bb0f4559786ef22c80496076bfe71638","abstract_canon_sha256":"01f3281bc7a1985164d7e0e10747a15bd2b660eb02cdf69a20296f5c4123a4b8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:01:07.805875Z","signature_b64":"zbcAnSApq9HsEC3lB7x384T0bPF5AdZ+/YdKd5nsF/UbbgJKK/62U8Re7bg0EFCp/55ccJjzOR/9YJb+az0fDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"97bd7e6e3ea9eb1bdffb8448aadcf54c53ef243c54d2772c68f9d00375a86745","last_reissued_at":"2026-07-05T11:01:07.805463Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:01:07.805463Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding and Mitigating Toxicity in Image-Text Pretraining Datasets: A Case Study on LLaVA","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Karthik Reddy Kanjula, Nahid Alam, Shayekh Bin Islam, Surya Guthikonda","submitted_at":"2025-05-09T18:01:50Z","abstract_excerpt":"Pretraining datasets are foundational to the development of multimodal models, yet they often have inherent biases and toxic content from the web-scale corpora they are sourced from. In this paper, we investigate the prevalence of toxicity in LLaVA image-text pretraining dataset, examining how harmful content manifests in different modalities. We present a comprehensive analysis of common toxicity categories and propose targeted mitigation strategies, resulting in the creation of a refined toxicity-mitigated dataset. This dataset removes 7,531 of toxic image-text pairs in the LLaVA pre-trainin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.06356","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/2505.06356/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":"2505.06356","created_at":"2026-07-05T11:01:07.805522+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.06356v1","created_at":"2026-07-05T11:01:07.805522+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.06356","created_at":"2026-07-05T11:01:07.805522+00:00"},{"alias_kind":"pith_short_12","alias_value":"S66X43R6VHVR","created_at":"2026-07-05T11:01:07.805522+00:00"},{"alias_kind":"pith_short_16","alias_value":"S66X43R6VHVRXX73","created_at":"2026-07-05T11:01:07.805522+00:00"},{"alias_kind":"pith_short_8","alias_value":"S66X43R6","created_at":"2026-07-05T11:01:07.805522+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR","json":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR.json","graph_json":"https://pith.science/api/pith-number/S66X43R6VHVRXX73QREKVXHVJR/graph.json","events_json":"https://pith.science/api/pith-number/S66X43R6VHVRXX73QREKVXHVJR/events.json","paper":"https://pith.science/paper/S66X43R6"},"agent_actions":{"view_html":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR","download_json":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR.json","view_paper":"https://pith.science/paper/S66X43R6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.06356&json=true","fetch_graph":"https://pith.science/api/pith-number/S66X43R6VHVRXX73QREKVXHVJR/graph.json","fetch_events":"https://pith.science/api/pith-number/S66X43R6VHVRXX73QREKVXHVJR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR/action/storage_attestation","attest_author":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR/action/author_attestation","sign_citation":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR/action/citation_signature","submit_replication":"https://pith.science/pith/S66X43R6VHVRXX73QREKVXHVJR/action/replication_record"}},"created_at":"2026-07-05T11:01:07.805522+00:00","updated_at":"2026-07-05T11:01:07.805522+00:00"}