{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:643I5SYCJWFWWV2KQV6JSH6LRT","short_pith_number":"pith:643I5SYC","schema_version":"1.0","canonical_sha256":"f7368ecb024d8b6b574a857c991fcb8cfa4e849032f9289ca95e01922931dc69","source":{"kind":"arxiv","id":"2505.23790","version":1},"attestation_state":"computed","paper":{"title":"Rethinking the Understanding Ability across LLMs through Mutual Information","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Na Zou, Shaojie Wang, Sirui Ding","submitted_at":"2025-05-25T22:31:24Z","abstract_excerpt":"Recent advances in large language models (LLMs) have revolutionized natural language processing, yet evaluating their intrinsic linguistic understanding remains challenging. Moving beyond specialized evaluation tasks, we propose an information-theoretic framework grounded in mutual information (MI) to achieve this. We formalize the understanding as MI between an input sentence and its latent representation (sentence-level MI), measuring how effectively input information is preserved in latent representation. Given that LLMs learn embeddings for individual tokens, we decompose sentence-level MI"},"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":"2505.23790","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-25T22:31:24Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"70ddad9175fb284c937b3aee8acf7a74ccd540ee7bbf6c3d5051b86dd80c4d23","abstract_canon_sha256":"031f7fefcd9d935cedc5752de7916eb1b9bdb005722e6aa8704f9c46e190dd11"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:12:35.376487Z","signature_b64":"n1RytQDaqJ459I0kBi79PCXzI6TyMvtuK2rwoaHNYz3UixB7gcOJX18oi+Az6nh5ge6hKhRGA+VIOOLlOM3zAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f7368ecb024d8b6b574a857c991fcb8cfa4e849032f9289ca95e01922931dc69","last_reissued_at":"2026-07-05T11:12:35.375933Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:12:35.375933Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rethinking the Understanding Ability across LLMs through Mutual Information","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Na Zou, Shaojie Wang, Sirui Ding","submitted_at":"2025-05-25T22:31:24Z","abstract_excerpt":"Recent advances in large language models (LLMs) have revolutionized natural language processing, yet evaluating their intrinsic linguistic understanding remains challenging. Moving beyond specialized evaluation tasks, we propose an information-theoretic framework grounded in mutual information (MI) to achieve this. We formalize the understanding as MI between an input sentence and its latent representation (sentence-level MI), measuring how effectively input information is preserved in latent representation. Given that LLMs learn embeddings for individual tokens, we decompose sentence-level MI"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.23790","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/2505.23790/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":"2505.23790","created_at":"2026-07-05T11:12:35.376015+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.23790v1","created_at":"2026-07-05T11:12:35.376015+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.23790","created_at":"2026-07-05T11:12:35.376015+00:00"},{"alias_kind":"pith_short_12","alias_value":"643I5SYCJWFW","created_at":"2026-07-05T11:12:35.376015+00:00"},{"alias_kind":"pith_short_16","alias_value":"643I5SYCJWFWWV2K","created_at":"2026-07-05T11:12:35.376015+00:00"},{"alias_kind":"pith_short_8","alias_value":"643I5SYC","created_at":"2026-07-05T11:12:35.376015+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/643I5SYCJWFWWV2KQV6JSH6LRT","json":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT.json","graph_json":"https://pith.science/api/pith-number/643I5SYCJWFWWV2KQV6JSH6LRT/graph.json","events_json":"https://pith.science/api/pith-number/643I5SYCJWFWWV2KQV6JSH6LRT/events.json","paper":"https://pith.science/paper/643I5SYC"},"agent_actions":{"view_html":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT","download_json":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT.json","view_paper":"https://pith.science/paper/643I5SYC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.23790&json=true","fetch_graph":"https://pith.science/api/pith-number/643I5SYCJWFWWV2KQV6JSH6LRT/graph.json","fetch_events":"https://pith.science/api/pith-number/643I5SYCJWFWWV2KQV6JSH6LRT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT/action/storage_attestation","attest_author":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT/action/author_attestation","sign_citation":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT/action/citation_signature","submit_replication":"https://pith.science/pith/643I5SYCJWFWWV2KQV6JSH6LRT/action/replication_record"}},"created_at":"2026-07-05T11:12:35.376015+00:00","updated_at":"2026-07-05T11:12:35.376015+00:00"}