{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7H37MRBVWO67NCLVQD5OG4EJAF","short_pith_number":"pith:7H37MRBV","schema_version":"1.0","canonical_sha256":"f9f7f64435b3bdf6897580fae37089014220fec3e558866a5f586c0210579c05","source":{"kind":"arxiv","id":"2607.03929","version":1},"attestation_state":"computed","paper":{"title":"Probe, Don't Prompt: A Hidden-State Probe for Metadata Filtering in Multi-Meta-RAG","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Mykhailo Poliakov, Nadiya Shvai","submitted_at":"2026-07-04T15:47:38Z","abstract_excerpt":"Multi-Meta-RAG improves retrieval for multi-hop question answering by filtering a vector store on metadata (the news source) that it extracts from each query by prompting gpt-3.5-turbo. We show this proprietary, free-form extractor can be replaced by a local, deterministic probe trained on the hidden states of a small open-source language model. On all 2556 MultiHop-RAG queries the probe reaches 90.9% set-exact accuracy against 88.0% for a model-free substring baseline and 80.9% for GPT-3.5, a margin that comes entirely from null queries, on which GPT-3.5 never abstains; on non-null queries al"},"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":"2607.03929","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-04T15:47:38Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"39d0614affcdd18576d602fa19c41dd1009b898738126b2a6ce32e837788e411","abstract_canon_sha256":"080e451126cd46cdc28b6350779ee5876a396478e4beb5b7d06395b932f388ce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:18:15.116004Z","signature_b64":"BmnDBpE0/GNVjpIc/826NyGHCAs7T370+uH79md3jUsYuUinAPNq/MLC77qENbUVWMgCLw1qXq2KIahcLpSNBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f9f7f64435b3bdf6897580fae37089014220fec3e558866a5f586c0210579c05","last_reissued_at":"2026-07-07T02:18:15.115210Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:18:15.115210Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Probe, Don't Prompt: A Hidden-State Probe for Metadata Filtering in Multi-Meta-RAG","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Mykhailo Poliakov, Nadiya Shvai","submitted_at":"2026-07-04T15:47:38Z","abstract_excerpt":"Multi-Meta-RAG improves retrieval for multi-hop question answering by filtering a vector store on metadata (the news source) that it extracts from each query by prompting gpt-3.5-turbo. We show this proprietary, free-form extractor can be replaced by a local, deterministic probe trained on the hidden states of a small open-source language model. On all 2556 MultiHop-RAG queries the probe reaches 90.9% set-exact accuracy against 88.0% for a model-free substring baseline and 80.9% for GPT-3.5, a margin that comes entirely from null queries, on which GPT-3.5 never abstains; on non-null queries al"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.03929","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/2607.03929/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":"2607.03929","created_at":"2026-07-07T02:18:15.115326+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.03929v1","created_at":"2026-07-07T02:18:15.115326+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.03929","created_at":"2026-07-07T02:18:15.115326+00:00"},{"alias_kind":"pith_short_12","alias_value":"7H37MRBVWO67","created_at":"2026-07-07T02:18:15.115326+00:00"},{"alias_kind":"pith_short_16","alias_value":"7H37MRBVWO67NCLV","created_at":"2026-07-07T02:18:15.115326+00:00"},{"alias_kind":"pith_short_8","alias_value":"7H37MRBV","created_at":"2026-07-07T02:18:15.115326+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/7H37MRBVWO67NCLVQD5OG4EJAF","json":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF.json","graph_json":"https://pith.science/api/pith-number/7H37MRBVWO67NCLVQD5OG4EJAF/graph.json","events_json":"https://pith.science/api/pith-number/7H37MRBVWO67NCLVQD5OG4EJAF/events.json","paper":"https://pith.science/paper/7H37MRBV"},"agent_actions":{"view_html":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF","download_json":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF.json","view_paper":"https://pith.science/paper/7H37MRBV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.03929&json=true","fetch_graph":"https://pith.science/api/pith-number/7H37MRBVWO67NCLVQD5OG4EJAF/graph.json","fetch_events":"https://pith.science/api/pith-number/7H37MRBVWO67NCLVQD5OG4EJAF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF/action/storage_attestation","attest_author":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF/action/author_attestation","sign_citation":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF/action/citation_signature","submit_replication":"https://pith.science/pith/7H37MRBVWO67NCLVQD5OG4EJAF/action/replication_record"}},"created_at":"2026-07-07T02:18:15.115326+00:00","updated_at":"2026-07-07T02:18:15.115326+00:00"}