{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7WAP2BCD5DWZUHYWZL33DV5NYT","short_pith_number":"pith:7WAP2BCD","schema_version":"1.0","canonical_sha256":"fd80fd0443e8ed9a1f16caf7b1d7adc4f58228907e944fd4dc5769fc2f0065de","source":{"kind":"arxiv","id":"2607.14306","version":1},"attestation_state":"computed","paper":{"title":"Tracing LLM Behavior to the Training Data with Empirical Next-Token Distributions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Zachary Izzo","submitted_at":"2026-07-15T19:11:54Z","abstract_excerpt":"In this paper, we study the connection between an LLM's output distribution and the data used to train it. Specifically, we study the degree to which an LLM's next-token distribution agrees with the empirical next-token distribution (ENTD) given the context in the training data. The ENTD is an appealing target because it is the unrestricted global minimizer of the next-token cross entropy loss used for pretraining, as well as an easily interpretable function of the pretraining corpus. We find that for a significant fraction of inputs, the LLM's distribution agrees with the ENTD almost perfectl"},"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":"2607.14306","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-07-15T19:11:54Z","cross_cats_sorted":[],"title_canon_sha256":"5a69019ab796b096b9779b3cccd7b9a299c7ade8bc43dc79e77ed40effc7044a","abstract_canon_sha256":"2dcc39dea006737960ce37a18ff3da283e6620580d3535837861f0195a9d437e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-17T00:21:04.439827Z","signature_b64":"3F6EC32DGy9XwTAGvpCNJblCrYGY3Q4kw536mD1uTpETkChNRmo8ce5SXGJy12neCCfmoQtlOjiryTmwtWp9Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fd80fd0443e8ed9a1f16caf7b1d7adc4f58228907e944fd4dc5769fc2f0065de","last_reissued_at":"2026-07-17T00:21:04.438946Z","signature_status":"signed_v1","first_computed_at":"2026-07-17T00:21:04.438946Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tracing LLM Behavior to the Training Data with Empirical Next-Token Distributions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Zachary Izzo","submitted_at":"2026-07-15T19:11:54Z","abstract_excerpt":"In this paper, we study the connection between an LLM's output distribution and the data used to train it. Specifically, we study the degree to which an LLM's next-token distribution agrees with the empirical next-token distribution (ENTD) given the context in the training data. The ENTD is an appealing target because it is the unrestricted global minimizer of the next-token cross entropy loss used for pretraining, as well as an easily interpretable function of the pretraining corpus. We find that for a significant fraction of inputs, the LLM's distribution agrees with the ENTD almost perfectl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.14306","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/2607.14306/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":"2607.14306","created_at":"2026-07-17T00:21:04.439413+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.14306v1","created_at":"2026-07-17T00:21:04.439413+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.14306","created_at":"2026-07-17T00:21:04.439413+00:00"},{"alias_kind":"pith_short_12","alias_value":"7WAP2BCD5DWZ","created_at":"2026-07-17T00:21:04.439413+00:00"},{"alias_kind":"pith_short_16","alias_value":"7WAP2BCD5DWZUHYW","created_at":"2026-07-17T00:21:04.439413+00:00"},{"alias_kind":"pith_short_8","alias_value":"7WAP2BCD","created_at":"2026-07-17T00:21:04.439413+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/7WAP2BCD5DWZUHYWZL33DV5NYT","json":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT.json","graph_json":"https://pith.science/api/pith-number/7WAP2BCD5DWZUHYWZL33DV5NYT/graph.json","events_json":"https://pith.science/api/pith-number/7WAP2BCD5DWZUHYWZL33DV5NYT/events.json","paper":"https://pith.science/paper/7WAP2BCD"},"agent_actions":{"view_html":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT","download_json":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT.json","view_paper":"https://pith.science/paper/7WAP2BCD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.14306&json=true","fetch_graph":"https://pith.science/api/pith-number/7WAP2BCD5DWZUHYWZL33DV5NYT/graph.json","fetch_events":"https://pith.science/api/pith-number/7WAP2BCD5DWZUHYWZL33DV5NYT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT/action/storage_attestation","attest_author":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT/action/author_attestation","sign_citation":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT/action/citation_signature","submit_replication":"https://pith.science/pith/7WAP2BCD5DWZUHYWZL33DV5NYT/action/replication_record"}},"created_at":"2026-07-17T00:21:04.439413+00:00","updated_at":"2026-07-17T00:21:04.439413+00:00"}