{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LUTV4YT4O7GY4OEPQV7UDYHVLP","short_pith_number":"pith:LUTV4YT4","schema_version":"1.0","canonical_sha256":"5d275e627c77cd8e388f857f41e0f55bd4f3b52d13134ad0f2e01ad706b329fa","source":{"kind":"arxiv","id":"2607.22646","version":1},"attestation_state":"computed","paper":{"title":"Extracting Algorithms in Pre-trained LLMs: A Case on Hidden Markov Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Jennifer J. Sun, Sarah Dean, Yahya Sattar, Yijia Dai, Zhaolin Gao","submitted_at":"2026-06-24T17:42:19Z","abstract_excerpt":"Large language models (LLMs) display a striking ability to predict next observations from Hidden Markov Models (HMMs) via in-context learning (ICL), but the algorithm underlying this capability remains undetermined: prior work has proposed several candidates without consensus, and none has been grounded in the model's internal activations. We close this gap with a three-stage pipeline. First, we empirically compare LLM behavior against a suite of candidate algorithms and narrow the space to three classes -- though no single class explains LLM behavior across all HMM settings and sequence lengt"},"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":"2607.22646","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-24T17:42:19Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"418474d4da1cf3436075d76c499af0853c0766259a9ab2ffea80a9cd449ad5a8","abstract_canon_sha256":"376d30a24dbdf443fca3785eb694175eeb60fd30a2cbb24cb360a093f3b88189"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-28T00:21:47.010202Z","signature_b64":"b6PbnVjfAXyeX+pdUJ30bXApXz9uvq3+TSGF58Zsx1JESM5D0Y3xMeBAIVl337DZvARMsk0q5L0BpsrSuG4qCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5d275e627c77cd8e388f857f41e0f55bd4f3b52d13134ad0f2e01ad706b329fa","last_reissued_at":"2026-07-28T00:21:47.009327Z","signature_status":"signed_v1","first_computed_at":"2026-07-28T00:21:47.009327Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Extracting Algorithms in Pre-trained LLMs: A Case on Hidden Markov Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.AI","authors_text":"Jennifer J. Sun, Sarah Dean, Yahya Sattar, Yijia Dai, Zhaolin Gao","submitted_at":"2026-06-24T17:42:19Z","abstract_excerpt":"Large language models (LLMs) display a striking ability to predict next observations from Hidden Markov Models (HMMs) via in-context learning (ICL), but the algorithm underlying this capability remains undetermined: prior work has proposed several candidates without consensus, and none has been grounded in the model's internal activations. We close this gap with a three-stage pipeline. First, we empirically compare LLM behavior against a suite of candidate algorithms and narrow the space to three classes -- though no single class explains LLM behavior across all HMM settings and sequence lengt"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.22646","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/2607.22646/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":"2607.22646","created_at":"2026-07-28T00:21:47.009773+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.22646v1","created_at":"2026-07-28T00:21:47.009773+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.22646","created_at":"2026-07-28T00:21:47.009773+00:00"},{"alias_kind":"pith_short_12","alias_value":"LUTV4YT4O7GY","created_at":"2026-07-28T00:21:47.009773+00:00"},{"alias_kind":"pith_short_16","alias_value":"LUTV4YT4O7GY4OEP","created_at":"2026-07-28T00:21:47.009773+00:00"},{"alias_kind":"pith_short_8","alias_value":"LUTV4YT4","created_at":"2026-07-28T00:21:47.009773+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/LUTV4YT4O7GY4OEPQV7UDYHVLP","json":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP.json","graph_json":"https://pith.science/api/pith-number/LUTV4YT4O7GY4OEPQV7UDYHVLP/graph.json","events_json":"https://pith.science/api/pith-number/LUTV4YT4O7GY4OEPQV7UDYHVLP/events.json","paper":"https://pith.science/paper/LUTV4YT4"},"agent_actions":{"view_html":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP","download_json":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP.json","view_paper":"https://pith.science/paper/LUTV4YT4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.22646&json=true","fetch_graph":"https://pith.science/api/pith-number/LUTV4YT4O7GY4OEPQV7UDYHVLP/graph.json","fetch_events":"https://pith.science/api/pith-number/LUTV4YT4O7GY4OEPQV7UDYHVLP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP/action/storage_attestation","attest_author":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP/action/author_attestation","sign_citation":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP/action/citation_signature","submit_replication":"https://pith.science/pith/LUTV4YT4O7GY4OEPQV7UDYHVLP/action/replication_record"}},"created_at":"2026-07-28T00:21:47.009773+00:00","updated_at":"2026-07-28T00:21:47.009773+00:00"}