{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:PRDJUS7HPCJOHJQ4WJL3OCDOLR","short_pith_number":"pith:PRDJUS7H","schema_version":"1.0","canonical_sha256":"7c469a4be77892e3a61cb257b7086e5c5127c1f2f0d88c201e53bb5a381d0b20","source":{"kind":"arxiv","id":"2308.04430","version":2},"attestation_state":"computed","paper":{"title":"SILO Language Models: Isolating Legal Risk In a Nonparametric Datastore","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Eric Wallace, Hannaneh Hajishirzi, Luke Zettlemoyer, Noah A. Smith, Sewon Min, Suchin Gururangan, Weijia Shi","submitted_at":"2023-08-08T17:58:15Z","abstract_excerpt":"The legality of training language models (LMs) on copyrighted or otherwise restricted data is under intense debate. However, as we show, model performance significantly degrades if trained only on low-risk text (e.g., out-of-copyright books or government documents), due to its limited size and domain coverage. We present SILO, a new language model that manages this risk-performance tradeoff during inference. SILO is built by (1) training a parametric LM on Open License Corpus (OLC), a new corpus we curate with 228B tokens of public domain and permissively licensed text and (2) augmenting it wi"},"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":"2308.04430","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-08-08T17:58:15Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"efe84471e9087fa3d2f4e36a74508124e4c3aa51873ea86337bf98b0fb703471","abstract_canon_sha256":"7004eb3744ea820a27556dc262db7a8a27804a5b859221719745bc1811a14fa0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:50:24.063496Z","signature_b64":"GxdaDLyMlrIZJHNj8jwQBd/X/wP3f37tPVXMWDzIXB75W7LgOmRw/AhaOuE2tBtjN17MH+gbee8vtwe3t5PxDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7c469a4be77892e3a61cb257b7086e5c5127c1f2f0d88c201e53bb5a381d0b20","last_reissued_at":"2026-07-05T08:50:24.063080Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:50:24.063080Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SILO Language Models: Isolating Legal Risk In a Nonparametric Datastore","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Eric Wallace, Hannaneh Hajishirzi, Luke Zettlemoyer, Noah A. Smith, Sewon Min, Suchin Gururangan, Weijia Shi","submitted_at":"2023-08-08T17:58:15Z","abstract_excerpt":"The legality of training language models (LMs) on copyrighted or otherwise restricted data is under intense debate. However, as we show, model performance significantly degrades if trained only on low-risk text (e.g., out-of-copyright books or government documents), due to its limited size and domain coverage. We present SILO, a new language model that manages this risk-performance tradeoff during inference. SILO is built by (1) training a parametric LM on Open License Corpus (OLC), a new corpus we curate with 228B tokens of public domain and permissively licensed text and (2) augmenting it wi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.04430","kind":"arxiv","version":2},"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/2308.04430/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":"2308.04430","created_at":"2026-07-05T08:50:24.063136+00:00"},{"alias_kind":"arxiv_version","alias_value":"2308.04430v2","created_at":"2026-07-05T08:50:24.063136+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.04430","created_at":"2026-07-05T08:50:24.063136+00:00"},{"alias_kind":"pith_short_12","alias_value":"PRDJUS7HPCJO","created_at":"2026-07-05T08:50:24.063136+00:00"},{"alias_kind":"pith_short_16","alias_value":"PRDJUS7HPCJOHJQ4","created_at":"2026-07-05T08:50:24.063136+00:00"},{"alias_kind":"pith_short_8","alias_value":"PRDJUS7H","created_at":"2026-07-05T08:50:24.063136+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2310.16789","citing_title":"Detecting Pretraining Data from Large Language Models","ref_index":89,"is_internal_anchor":false},{"citing_arxiv_id":"2311.05232","citing_title":"A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions","ref_index":224,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR","json":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR.json","graph_json":"https://pith.science/api/pith-number/PRDJUS7HPCJOHJQ4WJL3OCDOLR/graph.json","events_json":"https://pith.science/api/pith-number/PRDJUS7HPCJOHJQ4WJL3OCDOLR/events.json","paper":"https://pith.science/paper/PRDJUS7H"},"agent_actions":{"view_html":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR","download_json":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR.json","view_paper":"https://pith.science/paper/PRDJUS7H","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2308.04430&json=true","fetch_graph":"https://pith.science/api/pith-number/PRDJUS7HPCJOHJQ4WJL3OCDOLR/graph.json","fetch_events":"https://pith.science/api/pith-number/PRDJUS7HPCJOHJQ4WJL3OCDOLR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR/action/storage_attestation","attest_author":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR/action/author_attestation","sign_citation":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR/action/citation_signature","submit_replication":"https://pith.science/pith/PRDJUS7HPCJOHJQ4WJL3OCDOLR/action/replication_record"}},"created_at":"2026-07-05T08:50:24.063136+00:00","updated_at":"2026-07-05T08:50:24.063136+00:00"}