{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:VBZJVNLXAMEHPPTNBZOQLMXD62","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":"54df1d816637fa44e57f28f0bdefe99f2d7c224a023b4190ce459b540aa5b65b","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-13T02:31:08Z","title_canon_sha256":"c3c94e9f6bfef8c08458d25369d38dde1ab3b5ed742fde71d1db23572c9b01ab"},"schema_version":"1.0","source":{"id":"2405.07436","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.07436","created_at":"2026-07-05T08:18:22Z"},{"alias_kind":"arxiv_version","alias_value":"2405.07436v1","created_at":"2026-07-05T08:18:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.07436","created_at":"2026-07-05T08:18:22Z"},{"alias_kind":"pith_short_12","alias_value":"VBZJVNLXAMEH","created_at":"2026-07-05T08:18:22Z"},{"alias_kind":"pith_short_16","alias_value":"VBZJVNLXAMEHPPTN","created_at":"2026-07-05T08:18:22Z"},{"alias_kind":"pith_short_8","alias_value":"VBZJVNLX","created_at":"2026-07-05T08:18:22Z"}],"graph_snapshots":[{"event_id":"sha256:e2042ac0453359f502ba66581f84b9648015a3899b03dea0c22a7308ffc39b50","target":"graph","created_at":"2026-07-05T08:18:22Z","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"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2405.07436/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) perform well at a myriad of tasks, but explaining the processes behind this performance is a challenge. This paper investigates whether LLMs can give faithful high-level explanations of their own internal processes. To explore this, we introduce a dataset, ArticulateRules, of few-shot text-based classification tasks generated by simple rules. Each rule is associated with a simple natural-language explanation. We test whether models that have learned to classify inputs competently (both in- and out-of-distribution) are able to articulate freeform natural language ex","authors_text":"Bilal Chughtai, Dane Sherburn, Owain Evans","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-13T02:31:08Z","title":"Can Language Models Explain Their Own Classification Behavior?"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.07436","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":"sha256:80931bdfb85a716e8398db6c1a62c8e5ec2b2955fb7bf586c408dbde4b11a9f5","target":"record","created_at":"2026-07-05T08:18:22Z","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":"54df1d816637fa44e57f28f0bdefe99f2d7c224a023b4190ce459b540aa5b65b","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-05-13T02:31:08Z","title_canon_sha256":"c3c94e9f6bfef8c08458d25369d38dde1ab3b5ed742fde71d1db23572c9b01ab"},"schema_version":"1.0","source":{"id":"2405.07436","kind":"arxiv","version":1}},"canonical_sha256":"a8729ab577030877be6d0e5d05b2e3f6b182e2a41085122b067e675a716c85ae","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a8729ab577030877be6d0e5d05b2e3f6b182e2a41085122b067e675a716c85ae","first_computed_at":"2026-07-05T08:18:22.528810Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:18:22.528810Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"R0EbpIsILXmW+UVtNVNwm2wKSX9gfuU/debDtlAxKdu8k8B7ehEObIbhhk6q2t1MiqCjRlVGBE0BDS8sjcOIBw==","signature_status":"signed_v1","signed_at":"2026-07-05T08:18:22.529173Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.07436","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:80931bdfb85a716e8398db6c1a62c8e5ec2b2955fb7bf586c408dbde4b11a9f5","sha256:e2042ac0453359f502ba66581f84b9648015a3899b03dea0c22a7308ffc39b50"],"state_sha256":"2561eb6e9d0d661e892605eac2f6ea8c888f68ed7d51831fde3350e3d2860a12"}