{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:W3KFXE7KDHXICWL2WHI7KHEUSE","short_pith_number":"pith:W3KFXE7K","schema_version":"1.0","canonical_sha256":"b6d45b93ea19ee81597ab1d1f51c94911de752e90c74f0f08410be6783c6d390","source":{"kind":"arxiv","id":"2406.05113","version":3},"attestation_state":"computed","paper":{"title":"LlavaGuard: An Open VLM-based Framework for Safeguarding Vision Datasets and Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Felix Friedrich, Kristian Kersting, Lukas Helff, Manuel Brack, Patrick Schramowski","submitted_at":"2024-06-07T17:44:32Z","abstract_excerpt":"This paper introduces LlavaGuard, a suite of VLM-based vision safeguards that address the critical need for reliable guardrails in the era of large-scale data and models. To this end, we establish a novel open framework, describing a customizable safety taxonomy, data preprocessing, augmentation, and training setup. For teaching a VLM safeguard on safety, we further create a multimodal safety dataset with high-quality human expert annotations, where each image is labeled with a safety rating, category, and rationale. We also employ advanced augmentations to support context-specific assessments"},"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":"2406.05113","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-06-07T17:44:32Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"e83c1591cb4a59fbcd5eaad0c6622fcb8a7189aec4fc1ef33c7169d042eb3794","abstract_canon_sha256":"f429579a71c3dd8a598d99e60ecdccd440c3e26ce8167b5f4e0ffc9963be406d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:21.883519Z","signature_b64":"/PHCDcChxq7n8juCks7KgTUNvfYrOv7T+vvtawp/gqGcOyoSgRAoL1gGhnAmCjb+/xda0JqTH8c/z1kfKLkzAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b6d45b93ea19ee81597ab1d1f51c94911de752e90c74f0f08410be6783c6d390","last_reissued_at":"2026-07-05T11:17:21.883030Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:21.883030Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LlavaGuard: An Open VLM-based Framework for Safeguarding Vision Datasets and Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Felix Friedrich, Kristian Kersting, Lukas Helff, Manuel Brack, Patrick Schramowski","submitted_at":"2024-06-07T17:44:32Z","abstract_excerpt":"This paper introduces LlavaGuard, a suite of VLM-based vision safeguards that address the critical need for reliable guardrails in the era of large-scale data and models. To this end, we establish a novel open framework, describing a customizable safety taxonomy, data preprocessing, augmentation, and training setup. For teaching a VLM safeguard on safety, we further create a multimodal safety dataset with high-quality human expert annotations, where each image is labeled with a safety rating, category, and rationale. We also employ advanced augmentations to support context-specific assessments"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.05113","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/2406.05113/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":"2406.05113","created_at":"2026-07-05T11:17:21.883090+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.05113v3","created_at":"2026-07-05T11:17:21.883090+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.05113","created_at":"2026-07-05T11:17:21.883090+00:00"},{"alias_kind":"pith_short_12","alias_value":"W3KFXE7KDHXI","created_at":"2026-07-05T11:17:21.883090+00:00"},{"alias_kind":"pith_short_16","alias_value":"W3KFXE7KDHXICWL2","created_at":"2026-07-05T11:17:21.883090+00:00"},{"alias_kind":"pith_short_8","alias_value":"W3KFXE7K","created_at":"2026-07-05T11:17:21.883090+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.05910","citing_title":"PolicyShiftGuard: Benchmarking and Improving Policy-Adaptive Image Guardrails","ref_index":6,"is_internal_anchor":true},{"citing_arxiv_id":"2606.25034","citing_title":"Yuvion VL: A Multimodal Foundation Model for Adversarial Content and AI Safety","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.25034","citing_title":"Yuvion VL: A Multimodal Foundation Model for Adversarial Content and AI Safety","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01481","citing_title":"SafeGen-Bench: Benchmarking Safety in Image-Conditioned Text-to-Video Generation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17610","citing_title":"SafeLens: Deliberate and Efficient Video Guardrails with Fast-and-Slow Screening","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19190","citing_title":"Going PLACES: Participatory Localized Red Teaming for Text-to-Image Safety in the Global South","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12876","citing_title":"Certified Robustness under Heterogeneous Perturbations via Hybrid Randomized Smoothing","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07879","citing_title":"FlowGuard: Towards Lightweight In-Generation Safety Detection for Diffusion Models via Linear Latent Decoding","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08846","citing_title":"Dictionary-Aligned Concept Control for Safeguarding Multimodal LLMs","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07760","citing_title":"RuleSafe-VL: Evaluating Rule-Conditioned Decision Reasoning in Vision-Language Content Moderation","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE","json":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE.json","graph_json":"https://pith.science/api/pith-number/W3KFXE7KDHXICWL2WHI7KHEUSE/graph.json","events_json":"https://pith.science/api/pith-number/W3KFXE7KDHXICWL2WHI7KHEUSE/events.json","paper":"https://pith.science/paper/W3KFXE7K"},"agent_actions":{"view_html":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE","download_json":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE.json","view_paper":"https://pith.science/paper/W3KFXE7K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.05113&json=true","fetch_graph":"https://pith.science/api/pith-number/W3KFXE7KDHXICWL2WHI7KHEUSE/graph.json","fetch_events":"https://pith.science/api/pith-number/W3KFXE7KDHXICWL2WHI7KHEUSE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE/action/storage_attestation","attest_author":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE/action/author_attestation","sign_citation":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE/action/citation_signature","submit_replication":"https://pith.science/pith/W3KFXE7KDHXICWL2WHI7KHEUSE/action/replication_record"}},"created_at":"2026-07-05T11:17:21.883090+00:00","updated_at":"2026-07-05T11:17:21.883090+00:00"}