{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KVIJXRWYC4YVKNQ6DV3TU4TM25","short_pith_number":"pith:KVIJXRWY","schema_version":"1.0","canonical_sha256":"55509bc6d8173155361e1d773a726cd773ff395a25614118bc42b293a0113c08","source":{"kind":"arxiv","id":"2606.29571","version":1},"attestation_state":"computed","paper":{"title":"Anisotropy Decides Cosine vs. Rank Metrics for Text Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"V.S. Raghu Parupudi","submitted_at":"2026-06-28T19:24:59Z","abstract_excerpt":"The standard way to compare two text embeddings is cosine similarity. Scattered studies report that a different metric does better, but never pin down the geometric condition that decides when, or why. We settle both with a comprehensive empirical study: nineteen parameter-free similarity metrics on nineteen encoders, from compact sentence transformers up to seven-billion-parameter large language models, across seven datasets. The answer is geometric. When an encoder spreads its variance evenly across directions, cosine is the best parameter-free choice and no other metric helps by a usable ma"},"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":"2606.29571","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-28T19:24:59Z","cross_cats_sorted":[],"title_canon_sha256":"9ea2c9cc23da7d0d93f30d6362146db177d71829e46781358e95cab329408d98","abstract_canon_sha256":"255e7043dbb975511c0cba9f4ae4dfb1dd9a15b9c39261f1fa5f7505c21a83b6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:18:12.516643Z","signature_b64":"lgl9W19DfAybV9Q24QAqd1uZLzbJy2uJwDmEW1cAU3AZ+IGIjjC9wKgPNlPmdpoW39vk9ctBqAMtsCXpoFkUAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"55509bc6d8173155361e1d773a726cd773ff395a25614118bc42b293a0113c08","last_reissued_at":"2026-06-30T01:18:12.516152Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:18:12.516152Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Anisotropy Decides Cosine vs. Rank Metrics for Text Embeddings","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"V.S. Raghu Parupudi","submitted_at":"2026-06-28T19:24:59Z","abstract_excerpt":"The standard way to compare two text embeddings is cosine similarity. Scattered studies report that a different metric does better, but never pin down the geometric condition that decides when, or why. We settle both with a comprehensive empirical study: nineteen parameter-free similarity metrics on nineteen encoders, from compact sentence transformers up to seven-billion-parameter large language models, across seven datasets. The answer is geometric. When an encoder spreads its variance evenly across directions, cosine is the best parameter-free choice and no other metric helps by a usable ma"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.29571","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/2606.29571/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":"2606.29571","created_at":"2026-06-30T01:18:12.516232+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.29571v1","created_at":"2026-06-30T01:18:12.516232+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.29571","created_at":"2026-06-30T01:18:12.516232+00:00"},{"alias_kind":"pith_short_12","alias_value":"KVIJXRWYC4YV","created_at":"2026-06-30T01:18:12.516232+00:00"},{"alias_kind":"pith_short_16","alias_value":"KVIJXRWYC4YVKNQ6","created_at":"2026-06-30T01:18:12.516232+00:00"},{"alias_kind":"pith_short_8","alias_value":"KVIJXRWY","created_at":"2026-06-30T01:18:12.516232+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/KVIJXRWYC4YVKNQ6DV3TU4TM25","json":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25.json","graph_json":"https://pith.science/api/pith-number/KVIJXRWYC4YVKNQ6DV3TU4TM25/graph.json","events_json":"https://pith.science/api/pith-number/KVIJXRWYC4YVKNQ6DV3TU4TM25/events.json","paper":"https://pith.science/paper/KVIJXRWY"},"agent_actions":{"view_html":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25","download_json":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25.json","view_paper":"https://pith.science/paper/KVIJXRWY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.29571&json=true","fetch_graph":"https://pith.science/api/pith-number/KVIJXRWYC4YVKNQ6DV3TU4TM25/graph.json","fetch_events":"https://pith.science/api/pith-number/KVIJXRWYC4YVKNQ6DV3TU4TM25/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25/action/storage_attestation","attest_author":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25/action/author_attestation","sign_citation":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25/action/citation_signature","submit_replication":"https://pith.science/pith/KVIJXRWYC4YVKNQ6DV3TU4TM25/action/replication_record"}},"created_at":"2026-06-30T01:18:12.516232+00:00","updated_at":"2026-06-30T01:18:12.516232+00:00"}