{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5VFDH7O3EG5L5GU25YEIDBFDHN","short_pith_number":"pith:5VFDH7O3","schema_version":"1.0","canonical_sha256":"ed4a33fddb21babe9a9aee088184a33b4074fd139da262ced77e5b736aea71f8","source":{"kind":"arxiv","id":"2605.24981","version":1},"attestation_state":"computed","paper":{"title":"Large Language Model Selection with Limited Annotations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Andreas Kirsch, Nezihe Merve G\\\"urel, Patrik Okanovic, Torsten Hoefler, Yavuz Durmazkeser","submitted_at":"2026-05-24T10:18:51Z","abstract_excerpt":"Choosing a Large Language Model (LLM) for a given task requires comparing many strong candidates, yet standard evaluation relies on costly annotations over fixed evaluation sets. To address this challenge, we develop SELECT-LLM, the first framework for active model selection of LLMs. SELECT-LLM aims to find a small set of queries whose annotations are most informative for identifying the best LLM for a given task. To this end, we introduce a query selection rule based on expected information gain, computed from pairwise similarities between candidate model outputs. Because this rule only uses "},"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":"2605.24981","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-24T10:18:51Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"e782eba7e9dbd4b0f7737fcd585a4766700ef6e021e7456dcd77861bc6c102bc","abstract_canon_sha256":"dffefacbccf7f8e5072c9c0e71acd770c1322ae3b9192284d94685475ed70847"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:04:09.196158Z","signature_b64":"uHOTof7/sno+Z+HbJPBIw0ZLVLPDYqTUmECZ1xwfvjniMkBNnQ5rov5NSttSDD/i/QTfWWZXO7nVeKTIYOieDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ed4a33fddb21babe9a9aee088184a33b4074fd139da262ced77e5b736aea71f8","last_reissued_at":"2026-05-26T01:04:09.195484Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:04:09.195484Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Large Language Model Selection with Limited Annotations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Andreas Kirsch, Nezihe Merve G\\\"urel, Patrik Okanovic, Torsten Hoefler, Yavuz Durmazkeser","submitted_at":"2026-05-24T10:18:51Z","abstract_excerpt":"Choosing a Large Language Model (LLM) for a given task requires comparing many strong candidates, yet standard evaluation relies on costly annotations over fixed evaluation sets. To address this challenge, we develop SELECT-LLM, the first framework for active model selection of LLMs. SELECT-LLM aims to find a small set of queries whose annotations are most informative for identifying the best LLM for a given task. To this end, we introduce a query selection rule based on expected information gain, computed from pairwise similarities between candidate model outputs. Because this rule only uses "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.24981","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/2605.24981/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":"2605.24981","created_at":"2026-05-26T01:04:09.195613+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.24981v1","created_at":"2026-05-26T01:04:09.195613+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.24981","created_at":"2026-05-26T01:04:09.195613+00:00"},{"alias_kind":"pith_short_12","alias_value":"5VFDH7O3EG5L","created_at":"2026-05-26T01:04:09.195613+00:00"},{"alias_kind":"pith_short_16","alias_value":"5VFDH7O3EG5L5GU2","created_at":"2026-05-26T01:04:09.195613+00:00"},{"alias_kind":"pith_short_8","alias_value":"5VFDH7O3","created_at":"2026-05-26T01:04:09.195613+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/5VFDH7O3EG5L5GU25YEIDBFDHN","json":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN.json","graph_json":"https://pith.science/api/pith-number/5VFDH7O3EG5L5GU25YEIDBFDHN/graph.json","events_json":"https://pith.science/api/pith-number/5VFDH7O3EG5L5GU25YEIDBFDHN/events.json","paper":"https://pith.science/paper/5VFDH7O3"},"agent_actions":{"view_html":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN","download_json":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN.json","view_paper":"https://pith.science/paper/5VFDH7O3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.24981&json=true","fetch_graph":"https://pith.science/api/pith-number/5VFDH7O3EG5L5GU25YEIDBFDHN/graph.json","fetch_events":"https://pith.science/api/pith-number/5VFDH7O3EG5L5GU25YEIDBFDHN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN/action/storage_attestation","attest_author":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN/action/author_attestation","sign_citation":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN/action/citation_signature","submit_replication":"https://pith.science/pith/5VFDH7O3EG5L5GU25YEIDBFDHN/action/replication_record"}},"created_at":"2026-05-26T01:04:09.195613+00:00","updated_at":"2026-05-26T01:04:09.195613+00:00"}