{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:TZBVH2UCH57UFKYOX24RCL4VK2","short_pith_number":"pith:TZBVH2UC","canonical_record":{"source":{"id":"2605.15134","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T17:41:52Z","cross_cats_sorted":[],"title_canon_sha256":"e525bd4e7cf044c16fecf61182f97d46ba200441ae04e2b8b1138161726b9511","abstract_canon_sha256":"545b0dff904342ad808e2f41d6a1ec574f3e58016c2667e9567b51e7e3517f6b"},"schema_version":"1.0"},"canonical_sha256":"9e4353ea823f7f42ab0ebeb9112f9556b566ce5844e266b3081c4e6b79951434","source":{"kind":"arxiv","id":"2605.15134","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15134","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15134v1","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15134","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"pith_short_12","alias_value":"TZBVH2UCH57U","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"TZBVH2UCH57UFKYO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"TZBVH2UC","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:TZBVH2UCH57UFKYOX24RCL4VK2","target":"record","payload":{"canonical_record":{"source":{"id":"2605.15134","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T17:41:52Z","cross_cats_sorted":[],"title_canon_sha256":"e525bd4e7cf044c16fecf61182f97d46ba200441ae04e2b8b1138161726b9511","abstract_canon_sha256":"545b0dff904342ad808e2f41d6a1ec574f3e58016c2667e9567b51e7e3517f6b"},"schema_version":"1.0"},"canonical_sha256":"9e4353ea823f7f42ab0ebeb9112f9556b566ce5844e266b3081c4e6b79951434","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:54.021505Z","signature_b64":"EputUrJv/Iuli3IQ41hazMYODsjXSpm5psFG75FIeM65ySMSl4u6O6tIXn/e4F5vNGSp29XEm69su14c8CX0AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9e4353ea823f7f42ab0ebeb9112f9556b566ce5844e266b3081c4e6b79951434","last_reissued_at":"2026-05-17T23:38:54.020847Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:54.020847Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.15134","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"npIDjn4MXiVEQtXZQlzC487YRlB40+iWC4D8VxwNFcD9mdg8VZvytEhwapPfB1NGOudkwvsZU7oa3uftrinwBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-22T03:05:55.736755Z"},"content_sha256":"bffe3f403fca571c0a78d509ba128ebc0ed08e3c789f4d819d7ecbbeabe9e71f","schema_version":"1.0","event_id":"sha256:bffe3f403fca571c0a78d509ba128ebc0ed08e3c789f4d819d7ecbbeabe9e71f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:TZBVH2UCH57UFKYOX24RCL4VK2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Training ML Models with Predictable Failures","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Scott Niekum, Will Schwarzer","submitted_at":"2026-05-14T17:41:52Z","abstract_excerpt":"Estimating how often an ML model will fail at deployment scale is central to pre-deployment safety assessment, but a feasible evaluation set is rarely large enough to observe the failures that matter. Jones et al. (2025) address this by extrapolating from the largest k failure scores in an evaluation set to predict deployment-scale failure rates. We give a finite-k decomposition of this estimator's forecast error and show that it has a built-in bias toward over-prediction in the typical case, which is the safety-favorable direction. This bias is offset when the evaluation set misses a rare hig"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.15134","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":""},"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Le3AsuAal+7aQeYt3o9Q2pa+PdPg58R5sNkxGobTTkeS9BWkEFU8TyXXGd9vvuoua6zDZX5BxuCzkXdFu8DEBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-22T03:05:55.737127Z"},"content_sha256":"15dc760940e45922f225b82b07a4520de2d97464298fc3e487bf020a095ff6ae","schema_version":"1.0","event_id":"sha256:15dc760940e45922f225b82b07a4520de2d97464298fc3e487bf020a095ff6ae"},{"event_type":"integrity_finding","subject_pith_number":"pith:2026:TZBVH2UCH57UFKYOX24RCL4VK2","target":"integrity","payload":{"note":"DOI in the printed bibliography is fragmented by whitespace or line breaks. A longer candidate (10.18653/v1/2025.findings-emnlp.147.URL) was visible in the surrounding text but could not be confirmed against doi.org as printed.","snippet":"Zhifeng Jiang, Zhihua Jin, and Guoliang He. Prompt- Keeper: Safeguarding system prompts for LLMs. In Findings of the Association for Computational Lin- guistics: EMNLP 2025, pages 2712–2728, Suzhou, China, November 2025. Association for Com","arxiv_id":"2605.15134","detector":"doi_compliance","evidence":{"ref_index":21,"verdict_class":"incontrovertible","resolved_title":null,"printed_excerpt":"Zhifeng Jiang, Zhihua Jin, and Guoliang He. Prompt- Keeper: Safeguarding system prompts for LLMs. In Findings of the Association for Computational Lin- guistics: EMNLP 2025, pages 2712–2728, Suzhou, China, November 2025. Association for Com","reconstructed_doi":"10.18653/v1/2025.findings-emnlp.147.URL"},"severity":"advisory","ref_index":21,"audited_at":"2026-05-20T20:34:47.281837Z","event_type":"pith.integrity.v1","detected_doi":"10.18653/v1/2025.findings-emnlp.147.URL","detector_url":"https://pith.science/pith-integrity-protocol#doi_compliance","external_url":null,"finding_type":"recoverable_identifier","evidence_hash":"fa0fbccbb903e8818236fea4db63b26c0ec5d9885d2fda6b67c959c244300341","paper_version":2,"verdict_class":"incontrovertible","resolved_title":null,"detector_version":"1.0.0","detected_arxiv_id":null,"integrity_event_id":5443,"payload_sha256":"beccc3e2455db373ff453dd5ece81b3eebfae74ff928f719c1007232c7f5d05f","signature_b64":"TKFnWo5RkGPpLBV/G6h0eiKB6T6esBwZ4HhlztwMRvKXv9uHuMmh0kRvSQVEbx4uFxNeynL4ao+xipJKqqDNCg==","signing_key_id":"pith-v1-2026-05"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T20:38:31Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Cgp3sveDIPRyKFd384NcFsGmNyeXEKD8Y9DjzsCmXy7Rpx/nBvmxKxPEreWhA7kHW7y4o3p4WqpMYH93ErlYAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-22T03:05:55.738631Z"},"content_sha256":"6262ae0a6ee523c7697a38c48582caf5c3dde659dda1f7b6cf8b2218f8ed2de2","schema_version":"1.0","event_id":"sha256:6262ae0a6ee523c7697a38c48582caf5c3dde659dda1f7b6cf8b2218f8ed2de2"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TZBVH2UCH57UFKYOX24RCL4VK2/bundle.json","state_url":"https://pith.science/pith/TZBVH2UCH57UFKYOX24RCL4VK2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TZBVH2UCH57UFKYOX24RCL4VK2/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-22T03:05:55Z","links":{"resolver":"https://pith.science/pith/TZBVH2UCH57UFKYOX24RCL4VK2","bundle":"https://pith.science/pith/TZBVH2UCH57UFKYOX24RCL4VK2/bundle.json","state":"https://pith.science/pith/TZBVH2UCH57UFKYOX24RCL4VK2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TZBVH2UCH57UFKYOX24RCL4VK2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:TZBVH2UCH57UFKYOX24RCL4VK2","merge_version":"pith-open-graph-merge-v1","event_count":3,"valid_event_count":3,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"545b0dff904342ad808e2f41d6a1ec574f3e58016c2667e9567b51e7e3517f6b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T17:41:52Z","title_canon_sha256":"e525bd4e7cf044c16fecf61182f97d46ba200441ae04e2b8b1138161726b9511"},"schema_version":"1.0","source":{"id":"2605.15134","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15134","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15134v1","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15134","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"pith_short_12","alias_value":"TZBVH2UCH57U","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"TZBVH2UCH57UFKYO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"TZBVH2UC","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:15dc760940e45922f225b82b07a4520de2d97464298fc3e487bf020a095ff6ae","target":"graph","created_at":"2026-05-17T23:38:54Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Estimating how often an ML model will fail at deployment scale is central to pre-deployment safety assessment, but a feasible evaluation set is rarely large enough to observe the failures that matter. Jones et al. (2025) address this by extrapolating from the largest k failure scores in an evaluation set to predict deployment-scale failure rates. We give a finite-k decomposition of this estimator's forecast error and show that it has a built-in bias toward over-prediction in the typical case, which is the safety-favorable direction. This bias is offset when the evaluation set misses a rare hig","authors_text":"Scott Niekum, Will Schwarzer","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T17:41:52Z","title":"Training ML Models with Predictable Failures"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.15134","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:bffe3f403fca571c0a78d509ba128ebc0ed08e3c789f4d819d7ecbbeabe9e71f","target":"record","created_at":"2026-05-17T23:38:54Z","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":"545b0dff904342ad808e2f41d6a1ec574f3e58016c2667e9567b51e7e3517f6b","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T17:41:52Z","title_canon_sha256":"e525bd4e7cf044c16fecf61182f97d46ba200441ae04e2b8b1138161726b9511"},"schema_version":"1.0","source":{"id":"2605.15134","kind":"arxiv","version":1}},"canonical_sha256":"9e4353ea823f7f42ab0ebeb9112f9556b566ce5844e266b3081c4e6b79951434","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9e4353ea823f7f42ab0ebeb9112f9556b566ce5844e266b3081c4e6b79951434","first_computed_at":"2026-05-17T23:38:54.020847Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:54.020847Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"EputUrJv/Iuli3IQ41hazMYODsjXSpm5psFG75FIeM65ySMSl4u6O6tIXn/e4F5vNGSp29XEm69su14c8CX0AA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:54.021505Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.15134","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:bffe3f403fca571c0a78d509ba128ebc0ed08e3c789f4d819d7ecbbeabe9e71f","sha256:15dc760940e45922f225b82b07a4520de2d97464298fc3e487bf020a095ff6ae","sha256:6262ae0a6ee523c7697a38c48582caf5c3dde659dda1f7b6cf8b2218f8ed2de2"],"state_sha256":"cb225e6c938f886fb2fd63f5b6ec9b4dcdd4df203471ca481781b7170057b28a"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PRfcfJ3IcqRmcvuaGxy124WKT4RHQgA/QvRMTeJ6xs1EsQP3wqmnfK+qiI58XTNvcVwDbFfnjAl2Hpx9ZOJ8CA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-22T03:05:55.741034Z","bundle_sha256":"1fd8106dd67a7d4b36e84d212cdda14b383944186ac6c0a5b4198ac1da946020"}}