{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:IYW62XVBM67PIP2EAK2RDE7UNE","short_pith_number":"pith:IYW62XVB","schema_version":"1.0","canonical_sha256":"462ded5ea167bef43f4402b51193f46926e02448f3d174c98e3ef2730ae5b608","source":{"kind":"arxiv","id":"2605.23901","version":1},"attestation_state":"computed","paper":{"title":"LLMs as Noisy Channels: A Shannon Perspective on Model Capacity and Scaling Laws","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Chen Zheng, Deyi Liu, Jing Liu, Thomas Hartvigsen, Xu Ouyang, Yiyuan Ma, Yuan Yang, Yuhang Cai","submitted_at":"2026-05-22T17:59:38Z","abstract_excerpt":"Existing scaling laws for Large Language Models (LLMs), predominantly monotonic power laws, fail to explain emerging non-monotonic phenomena such as catastrophic overtraining and quantization-induced degradation, where performance deteriorates despite increased compute.\n  We propose the Shannon Scaling Law, a unified theoretical framework that models LLM training as information transmission over a noisy channel, grounded in the Shannon-Hartley theorem. By mapping model parameters to channel bandwidth and training tokens to signal power, our formulation explicitly captures the interaction betwe"},"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":"2605.23901","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-22T17:59:38Z","cross_cats_sorted":["cs.AI","cs.IT","math.IT"],"title_canon_sha256":"f525712475ccbc1ce28bcb2f30ee8edf58f6dde46a4ff56f5547c9074d427edf","abstract_canon_sha256":"f8e240b5753c71c325bbee8898973f1e015ba3a80f7b078c230d41760a232895"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-25T02:02:38.648767Z","signature_b64":"cMNiLMDyXxyHxkEl9RL+agviV7P6XA252jL5NRe0Xl4S+KErhxXIFXr6zV5bW26moWyL9JZsUJUO0gylVyVvDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"462ded5ea167bef43f4402b51193f46926e02448f3d174c98e3ef2730ae5b608","last_reissued_at":"2026-05-25T02:02:38.648048Z","signature_status":"signed_v1","first_computed_at":"2026-05-25T02:02:38.648048Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LLMs as Noisy Channels: A Shannon Perspective on Model Capacity and Scaling Laws","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Chen Zheng, Deyi Liu, Jing Liu, Thomas Hartvigsen, Xu Ouyang, Yiyuan Ma, Yuan Yang, Yuhang Cai","submitted_at":"2026-05-22T17:59:38Z","abstract_excerpt":"Existing scaling laws for Large Language Models (LLMs), predominantly monotonic power laws, fail to explain emerging non-monotonic phenomena such as catastrophic overtraining and quantization-induced degradation, where performance deteriorates despite increased compute.\n  We propose the Shannon Scaling Law, a unified theoretical framework that models LLM training as information transmission over a noisy channel, grounded in the Shannon-Hartley theorem. By mapping model parameters to channel bandwidth and training tokens to signal power, our formulation explicitly captures the interaction betwe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.23901","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/2605.23901/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":"2605.23901","created_at":"2026-05-25T02:02:38.648156+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.23901v1","created_at":"2026-05-25T02:02:38.648156+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.23901","created_at":"2026-05-25T02:02:38.648156+00:00"},{"alias_kind":"pith_short_12","alias_value":"IYW62XVBM67P","created_at":"2026-05-25T02:02:38.648156+00:00"},{"alias_kind":"pith_short_16","alias_value":"IYW62XVBM67PIP2E","created_at":"2026-05-25T02:02:38.648156+00:00"},{"alias_kind":"pith_short_8","alias_value":"IYW62XVB","created_at":"2026-05-25T02:02:38.648156+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/IYW62XVBM67PIP2EAK2RDE7UNE","json":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE.json","graph_json":"https://pith.science/api/pith-number/IYW62XVBM67PIP2EAK2RDE7UNE/graph.json","events_json":"https://pith.science/api/pith-number/IYW62XVBM67PIP2EAK2RDE7UNE/events.json","paper":"https://pith.science/paper/IYW62XVB"},"agent_actions":{"view_html":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE","download_json":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE.json","view_paper":"https://pith.science/paper/IYW62XVB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.23901&json=true","fetch_graph":"https://pith.science/api/pith-number/IYW62XVBM67PIP2EAK2RDE7UNE/graph.json","fetch_events":"https://pith.science/api/pith-number/IYW62XVBM67PIP2EAK2RDE7UNE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE/action/storage_attestation","attest_author":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE/action/author_attestation","sign_citation":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE/action/citation_signature","submit_replication":"https://pith.science/pith/IYW62XVBM67PIP2EAK2RDE7UNE/action/replication_record"}},"created_at":"2026-05-25T02:02:38.648156+00:00","updated_at":"2026-05-25T02:02:38.648156+00:00"}