{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QYP5OYO3DBZVBFJNBDX35ONZIC","short_pith_number":"pith:QYP5OYO3","schema_version":"1.0","canonical_sha256":"861fd761db187350952d08efbeb9b940a9fe9bd489456c020e409de8c4a0a227","source":{"kind":"arxiv","id":"2607.29378","version":1},"attestation_state":"computed","paper":{"title":"PTP: Previous-Token Prediction based LLM Inversion for Near-Exact Prompt Reconstruction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Amit Sethi, Atanu R Sinha, Nagasai Saketh Naidu, Pirzada Suhail","submitted_at":"2026-07-31T13:01:43Z","abstract_excerpt":"Large language models (LLMs) generate text by auto-regressively sampling the next token. This inherently leads to a many-to-many mapping between prompts and responses, complicating the task of inferring prompts from observed outputs. Prior work on LLM inversion frames prompt recovery as a semantic reconstruction task. They rely on fine-tuning pretrained sequence-to-sequence models on large external datasets--and requiring access to model weights or logits--to generate semantically plausible prompts. In contrast, we present a functional approach to inverting a given LLM in a black-box setting, "},"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":"2607.29378","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-31T13:01:43Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"79bf560d0e47a28ddee53f2ae1f08a1fadc8d2c4a561ede880093e37c04065cf","abstract_canon_sha256":"d03835a7ca000c9e8d73765ef8e6c7f6109651585e43500b98ad810873b86534"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-03T01:33:35.012653Z","signature_b64":"D934UXhhbi2YqfkuyO6D63F0dzDuLYugKspNPzhJx1T+le3HWr1X9vBmEDZMQ/Ymv05YrnBaydDdohF2IRU1CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"861fd761db187350952d08efbeb9b940a9fe9bd489456c020e409de8c4a0a227","last_reissued_at":"2026-08-03T01:33:35.010885Z","signature_status":"signed_v1","first_computed_at":"2026-08-03T01:33:35.010885Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PTP: Previous-Token Prediction based LLM Inversion for Near-Exact Prompt Reconstruction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Amit Sethi, Atanu R Sinha, Nagasai Saketh Naidu, Pirzada Suhail","submitted_at":"2026-07-31T13:01:43Z","abstract_excerpt":"Large language models (LLMs) generate text by auto-regressively sampling the next token. This inherently leads to a many-to-many mapping between prompts and responses, complicating the task of inferring prompts from observed outputs. Prior work on LLM inversion frames prompt recovery as a semantic reconstruction task. They rely on fine-tuning pretrained sequence-to-sequence models on large external datasets--and requiring access to model weights or logits--to generate semantically plausible prompts. In contrast, we present a functional approach to inverting a given LLM in a black-box setting, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.29378","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/2607.29378/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":"2607.29378","created_at":"2026-08-03T01:33:35.011897+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.29378v1","created_at":"2026-08-03T01:33:35.011897+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.29378","created_at":"2026-08-03T01:33:35.011897+00:00"},{"alias_kind":"pith_short_12","alias_value":"QYP5OYO3DBZV","created_at":"2026-08-03T01:33:35.011897+00:00"},{"alias_kind":"pith_short_16","alias_value":"QYP5OYO3DBZVBFJN","created_at":"2026-08-03T01:33:35.011897+00:00"},{"alias_kind":"pith_short_8","alias_value":"QYP5OYO3","created_at":"2026-08-03T01:33:35.011897+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/QYP5OYO3DBZVBFJNBDX35ONZIC","json":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC.json","graph_json":"https://pith.science/api/pith-number/QYP5OYO3DBZVBFJNBDX35ONZIC/graph.json","events_json":"https://pith.science/api/pith-number/QYP5OYO3DBZVBFJNBDX35ONZIC/events.json","paper":"https://pith.science/paper/QYP5OYO3"},"agent_actions":{"view_html":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC","download_json":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC.json","view_paper":"https://pith.science/paper/QYP5OYO3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.29378&json=true","fetch_graph":"https://pith.science/api/pith-number/QYP5OYO3DBZVBFJNBDX35ONZIC/graph.json","fetch_events":"https://pith.science/api/pith-number/QYP5OYO3DBZVBFJNBDX35ONZIC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC/action/storage_attestation","attest_author":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC/action/author_attestation","sign_citation":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC/action/citation_signature","submit_replication":"https://pith.science/pith/QYP5OYO3DBZVBFJNBDX35ONZIC/action/replication_record"}},"created_at":"2026-08-03T01:33:35.011897+00:00","updated_at":"2026-08-03T01:33:35.011897+00:00"}