{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JJEZSPBZMHOFENTYXJ73W2X6HP","short_pith_number":"pith:JJEZSPBZ","schema_version":"1.0","canonical_sha256":"4a49993c3961dc523678ba7fbb6afe3bfb3c35c6ecd933d013f729e29ea7908d","source":{"kind":"arxiv","id":"2502.14855","version":2},"attestation_state":"computed","paper":{"title":"Prompt-to-Leaderboard","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Anastasios N. Angelopoulos, Connor Chen, Evan Frick, Ion Stoica, Joseph Tennyson, Tianle Li, Wei-Lin Chiang","submitted_at":"2025-02-20T18:58:07Z","abstract_excerpt":"Large language model (LLM) evaluations typically rely on aggregated metrics like accuracy or human preference, averaging across users and prompts. This averaging obscures user- and prompt-specific variations in model performance. To address this, we propose Prompt-to-Leaderboard (P2L), a method that produces leaderboards specific to a prompt. The core idea is to train an LLM taking natural language prompts as input to output a vector of Bradley-Terry coefficients which are then used to predict the human preference vote. The resulting prompt-dependent leaderboards allow for unsupervised task-sp"},"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":"2502.14855","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-20T18:58:07Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"afb313a8b3bdf4091fe7b76dd9e42958c60704dff947817a1068ec72b2557e27","abstract_canon_sha256":"8db5ed68a50c0b1bb63358c9efb947eac840e2951aa9cf19d54bea470a342550"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:27:25.857514Z","signature_b64":"gxTJciNfNt/L1Twft6kQC8SI9eMNuhthbjnv5KuXpOYIEtDnvknDFVQn14X6d2IQ5VYSGqW97+y+P92+Tr0ABA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4a49993c3961dc523678ba7fbb6afe3bfb3c35c6ecd933d013f729e29ea7908d","last_reissued_at":"2026-07-05T10:27:25.857007Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:27:25.857007Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Prompt-to-Leaderboard","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Anastasios N. Angelopoulos, Connor Chen, Evan Frick, Ion Stoica, Joseph Tennyson, Tianle Li, Wei-Lin Chiang","submitted_at":"2025-02-20T18:58:07Z","abstract_excerpt":"Large language model (LLM) evaluations typically rely on aggregated metrics like accuracy or human preference, averaging across users and prompts. This averaging obscures user- and prompt-specific variations in model performance. To address this, we propose Prompt-to-Leaderboard (P2L), a method that produces leaderboards specific to a prompt. The core idea is to train an LLM taking natural language prompts as input to output a vector of Bradley-Terry coefficients which are then used to predict the human preference vote. The resulting prompt-dependent leaderboards allow for unsupervised task-sp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.14855","kind":"arxiv","version":2},"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/2502.14855/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":"2502.14855","created_at":"2026-07-05T10:27:25.857074+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.14855v2","created_at":"2026-07-05T10:27:25.857074+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.14855","created_at":"2026-07-05T10:27:25.857074+00:00"},{"alias_kind":"pith_short_12","alias_value":"JJEZSPBZMHOF","created_at":"2026-07-05T10:27:25.857074+00:00"},{"alias_kind":"pith_short_16","alias_value":"JJEZSPBZMHOFENTY","created_at":"2026-07-05T10:27:25.857074+00:00"},{"alias_kind":"pith_short_8","alias_value":"JJEZSPBZ","created_at":"2026-07-05T10:27:25.857074+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17628","citing_title":"OPD-Evolver: Cultivating Holistic Agent Evolver via On-Policy Distillation","ref_index":149,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07587","citing_title":"The Routing Plateau: Understanding and Breaking the Accuracy Limits of LLM Routers","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03426","citing_title":"Replacing Parameters with Preferences: Federated Alignment of Heterogeneous Vision-Language Models","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10805","citing_title":"Reasoning Is Not Free: Robust Adaptive Cost-Efficient Routing for LLM-as-a-Judge","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18576","citing_title":"Agentic Forecasting using Sequential Bayesian Updating of Linguistic Beliefs","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15022","citing_title":"Route to Rome Attack: Directing LLM Routers to Expensive Models via Adversarial Suffix Optimization","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP","json":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP.json","graph_json":"https://pith.science/api/pith-number/JJEZSPBZMHOFENTYXJ73W2X6HP/graph.json","events_json":"https://pith.science/api/pith-number/JJEZSPBZMHOFENTYXJ73W2X6HP/events.json","paper":"https://pith.science/paper/JJEZSPBZ"},"agent_actions":{"view_html":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP","download_json":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP.json","view_paper":"https://pith.science/paper/JJEZSPBZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.14855&json=true","fetch_graph":"https://pith.science/api/pith-number/JJEZSPBZMHOFENTYXJ73W2X6HP/graph.json","fetch_events":"https://pith.science/api/pith-number/JJEZSPBZMHOFENTYXJ73W2X6HP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP/action/storage_attestation","attest_author":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP/action/author_attestation","sign_citation":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP/action/citation_signature","submit_replication":"https://pith.science/pith/JJEZSPBZMHOFENTYXJ73W2X6HP/action/replication_record"}},"created_at":"2026-07-05T10:27:25.857074+00:00","updated_at":"2026-07-05T10:27:25.857074+00:00"}