{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:RONT2KKFZHU76B46F7T5UHUHEM","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":"439cecdcc9d89c3b7ad203b555265245bf7fa4646bf758b43ac25786662b6095","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2025-12-15T02:57:55Z","title_canon_sha256":"a2c143534b0c446e7849a4155294f2a6c053239c299f007027ad28a799469333"},"schema_version":"1.0","source":{"id":"2512.14754","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.14754","created_at":"2026-05-29T02:05:38Z"},{"alias_kind":"arxiv_version","alias_value":"2512.14754v3","created_at":"2026-05-29T02:05:38Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.14754","created_at":"2026-05-29T02:05:38Z"},{"alias_kind":"pith_short_12","alias_value":"RONT2KKFZHU7","created_at":"2026-05-29T02:05:38Z"},{"alias_kind":"pith_short_16","alias_value":"RONT2KKFZHU76B46","created_at":"2026-05-29T02:05:38Z"},{"alias_kind":"pith_short_8","alias_value":"RONT2KKF","created_at":"2026-05-29T02:05:38Z"}],"graph_snapshots":[{"event_id":"sha256:c174ff4beaadb1dccd685d6ed26ba68ee705cc85088f215729779c729e3af8e0","target":"graph","created_at":"2026-05-29T02:05:38Z","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":"Across 20 proprietary and 26 open-source LLMs, we find that current models exhibit substantial insufficiency in nuance-oriented reliability -- their performance can drop by up to 61.8% with nuanced prompt modifications."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The automated data-augmentation pipeline produces cousin prompts that preserve the original user intent without changing task difficulty or introducing new ambiguities."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"LLMs exhibit up to 61.8% performance drops on nuanced rephrasings of instruction-following tasks, revealing insufficient nuance-oriented reliability across 46 tested models."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Language models lose up to 61.8 percent instruction-following accuracy on prompts with subtle phrasing changes that preserve intent."}],"snapshot_sha256":"4abc4ab846f600c229f7e45277e4e0cbd75a50b3ecb1246b0f4b18984ab2efa9"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2512.14754/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Advanced LLMs have achieved near-ceiling instruction-following accuracy on benchmarks such as IFEval. However, these impressive scores do not necessarily translate to reliable services in real-world use, where users often vary their phrasing, contextual framing, and task formulations. In this paper, we study nuance-oriented reliability: whether models exhibit consistent competence across cousin prompts that convey analogous user intents but with subtle nuances. To quantify this, we introduce a new metric, reliable@k, and develop an automated pipeline that generates high-quality cousin prompts ","authors_text":"Chao Zhang, Han Qiu, Jianshuo Dong, Tao Wei, Yan Liu, Yutong Zhang, Zhenyu Zhong","cross_cats":["cs.AI","cs.CL"],"headline":"Language models lose up to 61.8 percent instruction-following accuracy on prompts with subtle phrasing changes that preserve intent.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2025-12-15T02:57:55Z","title":"Revisiting the Reliability of Language Models in Instruction-Following"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.14754","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-16T22:49:18.604668Z","id":"45fde6dd-8b81-4a6e-a272-618791a01207","model_set":{"reader":"grok-4.3"},"one_line_summary":"LLMs exhibit up to 61.8% performance drops on nuanced rephrasings of instruction-following tasks, revealing insufficient nuance-oriented reliability across 46 tested models.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Language models lose up to 61.8 percent instruction-following accuracy on prompts with subtle phrasing changes that preserve intent.","strongest_claim":"Across 20 proprietary and 26 open-source LLMs, we find that current models exhibit substantial insufficiency in nuance-oriented reliability -- their performance can drop by up to 61.8% with nuanced prompt modifications.","weakest_assumption":"The automated data-augmentation pipeline produces cousin prompts that preserve the original user intent without changing task difficulty or introducing new ambiguities."}},"verdict_id":"45fde6dd-8b81-4a6e-a272-618791a01207"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:7e125d78a546e86dc08f3e2adb43a34ce4fb8f7123e4df4893c0dd0fd95e5bc3","target":"record","created_at":"2026-05-29T02:05:38Z","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":"439cecdcc9d89c3b7ad203b555265245bf7fa4646bf758b43ac25786662b6095","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SE","submitted_at":"2025-12-15T02:57:55Z","title_canon_sha256":"a2c143534b0c446e7849a4155294f2a6c053239c299f007027ad28a799469333"},"schema_version":"1.0","source":{"id":"2512.14754","kind":"arxiv","version":3}},"canonical_sha256":"8b9b3d2945c9e9ff079e2fe7da1e87231b1956fdff36323a959ef49730d3f32b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8b9b3d2945c9e9ff079e2fe7da1e87231b1956fdff36323a959ef49730d3f32b","first_computed_at":"2026-05-29T02:05:38.447274Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-29T02:05:38.447274Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"EMllw0NrS8eCUcoTtrPYyyL3zZhdwGsFTBcmeZo/b/iA4nooU/L6jQcsvs9DoVuvIry6wg+KRDWFw3uV5EhpBg==","signature_status":"signed_v1","signed_at":"2026-05-29T02:05:38.448510Z","signed_message":"canonical_sha256_bytes"},"source_id":"2512.14754","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7e125d78a546e86dc08f3e2adb43a34ce4fb8f7123e4df4893c0dd0fd95e5bc3","sha256:c174ff4beaadb1dccd685d6ed26ba68ee705cc85088f215729779c729e3af8e0"],"state_sha256":"6cfb4ac4201c6d7ff69837fb5aeda1b3879fd80a6d1198cffd2ad2f8aaec423b"}