{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HKDJ3UUCP2N6MDNDPSO6JUYXIX","short_pith_number":"pith:HKDJ3UUC","schema_version":"1.0","canonical_sha256":"3a869dd2827e9be60da37c9de4d31745de963de819cbc691d837d55cd47a88e1","source":{"kind":"arxiv","id":"2605.25893","version":1},"attestation_state":"computed","paper":{"title":"$D^2$-Monitor: Dynamic Safety Monitoring for Diffusion LLMs via Hesitation-Aware Routing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Adel Bibi, Aoxi Liu, Baoyuan Wu, Guanzhe Hong, James Oldfield, Junchi Yu, Philip Torr, YuPeng Chen","submitted_at":"2026-05-25T14:22:21Z","abstract_excerpt":"Despite the emergence of diffusion large language models (D-LLMs) as an alternative to autoregressive large language models (AR-LLMs), safety monitoring for D-LLMs remains largely unexplored. Unlike AR-LLMs, D-LLMs generate text through a multi-step denoising process, exposing intermediate hidden representations that may contain safety-relevant information unavailable in standard single-step monitoring setups. Motivated by the suitability of lightweight probes for always-on monitoring, we analyze which trajectory-level signals best indicate when such probes are likely to struggle. We find that"},"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":"2605.25893","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-25T14:22:21Z","cross_cats_sorted":[],"title_canon_sha256":"76f78e910403e1aa6cf96468d35eb28ece0ae04d693b1f060125ea06cfde46d4","abstract_canon_sha256":"d4e7a3a2d8d136d4584d729ee3640b21951f72b5a46f4ad9da2bfac384f6c296"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:05:17.252653Z","signature_b64":"v93uhi50/1ieRksODuk5zNasNrBmN5mtqsnqdB+lC/gVePl+0BoRWx1LhkLkteoYsdF3yuSjachySrMAeEVJDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3a869dd2827e9be60da37c9de4d31745de963de819cbc691d837d55cd47a88e1","last_reissued_at":"2026-05-26T02:05:17.251965Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:05:17.251965Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"$D^2$-Monitor: Dynamic Safety Monitoring for Diffusion LLMs via Hesitation-Aware Routing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Adel Bibi, Aoxi Liu, Baoyuan Wu, Guanzhe Hong, James Oldfield, Junchi Yu, Philip Torr, YuPeng Chen","submitted_at":"2026-05-25T14:22:21Z","abstract_excerpt":"Despite the emergence of diffusion large language models (D-LLMs) as an alternative to autoregressive large language models (AR-LLMs), safety monitoring for D-LLMs remains largely unexplored. Unlike AR-LLMs, D-LLMs generate text through a multi-step denoising process, exposing intermediate hidden representations that may contain safety-relevant information unavailable in standard single-step monitoring setups. Motivated by the suitability of lightweight probes for always-on monitoring, we analyze which trajectory-level signals best indicate when such probes are likely to struggle. We find that"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.25893","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/2605.25893/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":"2605.25893","created_at":"2026-05-26T02:05:17.252069+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.25893v1","created_at":"2026-05-26T02:05:17.252069+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.25893","created_at":"2026-05-26T02:05:17.252069+00:00"},{"alias_kind":"pith_short_12","alias_value":"HKDJ3UUCP2N6","created_at":"2026-05-26T02:05:17.252069+00:00"},{"alias_kind":"pith_short_16","alias_value":"HKDJ3UUCP2N6MDND","created_at":"2026-05-26T02:05:17.252069+00:00"},{"alias_kind":"pith_short_8","alias_value":"HKDJ3UUC","created_at":"2026-05-26T02:05:17.252069+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/HKDJ3UUCP2N6MDNDPSO6JUYXIX","json":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX.json","graph_json":"https://pith.science/api/pith-number/HKDJ3UUCP2N6MDNDPSO6JUYXIX/graph.json","events_json":"https://pith.science/api/pith-number/HKDJ3UUCP2N6MDNDPSO6JUYXIX/events.json","paper":"https://pith.science/paper/HKDJ3UUC"},"agent_actions":{"view_html":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX","download_json":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX.json","view_paper":"https://pith.science/paper/HKDJ3UUC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.25893&json=true","fetch_graph":"https://pith.science/api/pith-number/HKDJ3UUCP2N6MDNDPSO6JUYXIX/graph.json","fetch_events":"https://pith.science/api/pith-number/HKDJ3UUCP2N6MDNDPSO6JUYXIX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX/action/storage_attestation","attest_author":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX/action/author_attestation","sign_citation":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX/action/citation_signature","submit_replication":"https://pith.science/pith/HKDJ3UUCP2N6MDNDPSO6JUYXIX/action/replication_record"}},"created_at":"2026-05-26T02:05:17.252069+00:00","updated_at":"2026-05-26T02:05:17.252069+00:00"}