{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:KA7ICOTDHZL4TITMUSLHMEBL3D","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"87fcc8f582da8f4dfb1a9eb1534ecb945a80f771a6a8095ecc6320410f5b1e7f","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2025-12-26T01:01:25Z","title_canon_sha256":"48f6abf74e09a01bb218402c0ee098c18be6a07cd82f4b547d10d39545aea293"},"schema_version":"1.0","source":{"id":"2512.21815","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.21815","created_at":"2026-05-26T01:03:21Z"},{"alias_kind":"arxiv_version","alias_value":"2512.21815v3","created_at":"2026-05-26T01:03:21Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.21815","created_at":"2026-05-26T01:03:21Z"},{"alias_kind":"pith_short_12","alias_value":"KA7ICOTDHZL4","created_at":"2026-05-26T01:03:21Z"},{"alias_kind":"pith_short_16","alias_value":"KA7ICOTDHZL4TITM","created_at":"2026-05-26T01:03:21Z"},{"alias_kind":"pith_short_8","alias_value":"KA7ICOTD","created_at":"2026-05-26T01:03:21Z"}],"graph_snapshots":[{"event_id":"sha256:86a42b18411cc5028a8dbf3c7ae271d0d784ba22074a53f257159a084da4718c","target":"graph","created_at":"2026-05-26T01:03:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"a small fraction (around 20%) of high-entropy tokens, in the evaluated representative open-source VLMs with diverse architectures, concentrates a disproportionate share of adversarial influence during autoregressive generation"},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That high-entropy tokens can be identified reliably during generation and that concentrating perturbations on them produces comparable semantic degradation to global attacks without requiring post-hoc selection or model-specific tuning that would invalidate the transferability claim."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"High-entropy tokens act as concentrated multimodal failure points in VLMs, enabling sparse Entropy-Guided Attacks that achieve 93-95% success and 30-38% harmful rates with cross-model transfer."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"A small share of high-entropy tokens during generation concentrates most adversarial influence in vision-language models."}],"snapshot_sha256":"1f8077993881cfb232972a684346efbb0dc5737bc45ee675bedbd820673b7b48"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b318806ae64168b412aa3aabbb7615d7f56ade2d02e8bd896fead889c1bbbaef"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2512.21815/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Vision-language models (VLMs) achieve remarkable performance but remain vulnerable to adversarial attacks. Entropy, as a measure of model uncertainty, is highly correlated with VLM reliability. While prior entropy-based attacks maximize uncertainty at all decoding steps, implicitly assuming that every token equally contributes to model instability, we reveal that a small fraction (around 20%) of high-entropy tokens, in the evaluated representative open-source VLMs with diverse architectures, concentrates a disproportionate share of adversarial influence during autoregressive generation. We dem","authors_text":"Jing Zhang, Jinhong Ni, Mengqi He, Shu Zou, Xin Shen, Xinyu Tian, Zhaoyuan Yang","cross_cats":["cs.LG"],"headline":"A small share of high-entropy tokens during generation concentrates most adversarial influence in vision-language models.","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2025-12-26T01:01:25Z","title":"High-Entropy Tokens as Multimodal Failure Points in Vision-Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.21815","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-16T19:26:18.378293Z","id":"ec215b08-5ced-4a3d-a80c-8d4baa9b633b","model_set":{"reader":"grok-4.3"},"one_line_summary":"High-entropy tokens act as concentrated multimodal failure points in VLMs, enabling sparse Entropy-Guided Attacks that achieve 93-95% success and 30-38% harmful rates with cross-model transfer.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"A small share of high-entropy tokens during generation concentrates most adversarial influence in vision-language models.","strongest_claim":"a small fraction (around 20%) of high-entropy tokens, in the evaluated representative open-source VLMs with diverse architectures, concentrates a disproportionate share of adversarial influence during autoregressive generation","weakest_assumption":"That high-entropy tokens can be identified reliably during generation and that concentrating perturbations on them produces comparable semantic degradation to global attacks without requiring post-hoc selection or model-specific tuning that would invalidate the transferability claim."}},"verdict_id":"ec215b08-5ced-4a3d-a80c-8d4baa9b633b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:3d39be98e0d05769cf8d89f55bf1062fad4cb0b3b913ccf451e3f851d1f3ecc0","target":"record","created_at":"2026-05-26T01:03:21Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"87fcc8f582da8f4dfb1a9eb1534ecb945a80f771a6a8095ecc6320410f5b1e7f","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2025-12-26T01:01:25Z","title_canon_sha256":"48f6abf74e09a01bb218402c0ee098c18be6a07cd82f4b547d10d39545aea293"},"schema_version":"1.0","source":{"id":"2512.21815","kind":"arxiv","version":3}},"canonical_sha256":"503e813a633e57c9a26ca49676102bd8f47736e4df6975fe74e80cc419cbfb1e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"503e813a633e57c9a26ca49676102bd8f47736e4df6975fe74e80cc419cbfb1e","first_computed_at":"2026-05-26T01:03:21.256922Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-26T01:03:21.256922Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"sUUHHuqXZzX/fRx7tA/543sZzUuioYa6M8szCh529g8uOH/PV87V8HAyctQv6mfOGrXYtE4klpORRaJslQZ+Dg==","signature_status":"signed_v1","signed_at":"2026-05-26T01:03:21.257956Z","signed_message":"canonical_sha256_bytes"},"source_id":"2512.21815","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3d39be98e0d05769cf8d89f55bf1062fad4cb0b3b913ccf451e3f851d1f3ecc0","sha256:86a42b18411cc5028a8dbf3c7ae271d0d784ba22074a53f257159a084da4718c"],"state_sha256":"5bfe1378217308f1bf2264ada75ebd98f3ac5af446a87b03f378164286a00606"}