{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:J3J5LDU24MGFETJTK2F5YSQMUU","short_pith_number":"pith:J3J5LDU2","schema_version":"1.0","canonical_sha256":"4ed3d58e9ae30c524d33568bdc4a0ca5059aec51fd27dda1172369d614c0a1d2","source":{"kind":"arxiv","id":"2502.20315","version":1},"attestation_state":"computed","paper":{"title":"LangProBe: a Language Programs Benchmark","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Arnav Singhvi, Dan Klein, Koushik Sen, Lakshya A Agrawal, Liheng Lai, Matei Zaharia, Michael J Ryan, Omar Khattab, Shangyin Tan","submitted_at":"2025-02-27T17:41:49Z","abstract_excerpt":"Composing language models (LMs) into multi-step language programs and automatically optimizing their modular prompts is now a mainstream paradigm for building AI systems, but the tradeoffs in this space have only scarcely been studied before. We introduce LangProBe, the first large-scale benchmark for evaluating the architectures and optimization strategies for language programs, with over 2000 combinations of tasks, architectures, optimizers, and choices of LMs. Using LangProBe, we are the first to study the impact of program architectures and optimizers (and their compositions together and w"},"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":"2502.20315","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-27T17:41:49Z","cross_cats_sorted":["cs.AI","cs.IR","cs.LG"],"title_canon_sha256":"8a0926ee7ed1c2a49d1996b72c86ea14c5cc2af56c541aed20c2bed58405a4bb","abstract_canon_sha256":"ca33f33948b0d49940a027a05c3735cf31b91d0f98c8b1849cec02ebf5ffec74"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:21:12.906696Z","signature_b64":"fgxxiXgx9KPJwGUmWN7EPXiUnR4WViiOHniQywC9JRdu5ZocTqtQnmrIKnaOsugnWuu+aAlegR77D91Liq8lBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4ed3d58e9ae30c524d33568bdc4a0ca5059aec51fd27dda1172369d614c0a1d2","last_reissued_at":"2026-07-05T10:21:12.906179Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:21:12.906179Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LangProBe: a Language Programs Benchmark","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Arnav Singhvi, Dan Klein, Koushik Sen, Lakshya A Agrawal, Liheng Lai, Matei Zaharia, Michael J Ryan, Omar Khattab, Shangyin Tan","submitted_at":"2025-02-27T17:41:49Z","abstract_excerpt":"Composing language models (LMs) into multi-step language programs and automatically optimizing their modular prompts is now a mainstream paradigm for building AI systems, but the tradeoffs in this space have only scarcely been studied before. We introduce LangProBe, the first large-scale benchmark for evaluating the architectures and optimization strategies for language programs, with over 2000 combinations of tasks, architectures, optimizers, and choices of LMs. Using LangProBe, we are the first to study the impact of program architectures and optimizers (and their compositions together and w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.20315","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/2502.20315/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":"2502.20315","created_at":"2026-07-05T10:21:12.906245+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.20315v1","created_at":"2026-07-05T10:21:12.906245+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.20315","created_at":"2026-07-05T10:21:12.906245+00:00"},{"alias_kind":"pith_short_12","alias_value":"J3J5LDU24MGF","created_at":"2026-07-05T10:21:12.906245+00:00"},{"alias_kind":"pith_short_16","alias_value":"J3J5LDU24MGFETJT","created_at":"2026-07-05T10:21:12.906245+00:00"},{"alias_kind":"pith_short_8","alias_value":"J3J5LDU2","created_at":"2026-07-05T10:21:12.906245+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.03620","citing_title":"Is It Time To Treat Prompts As Code? A Multi-Use Case Study For Prompt Optimization Using DSPy","ref_index":16,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU","json":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU.json","graph_json":"https://pith.science/api/pith-number/J3J5LDU24MGFETJTK2F5YSQMUU/graph.json","events_json":"https://pith.science/api/pith-number/J3J5LDU24MGFETJTK2F5YSQMUU/events.json","paper":"https://pith.science/paper/J3J5LDU2"},"agent_actions":{"view_html":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU","download_json":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU.json","view_paper":"https://pith.science/paper/J3J5LDU2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.20315&json=true","fetch_graph":"https://pith.science/api/pith-number/J3J5LDU24MGFETJTK2F5YSQMUU/graph.json","fetch_events":"https://pith.science/api/pith-number/J3J5LDU24MGFETJTK2F5YSQMUU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU/action/storage_attestation","attest_author":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU/action/author_attestation","sign_citation":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU/action/citation_signature","submit_replication":"https://pith.science/pith/J3J5LDU24MGFETJTK2F5YSQMUU/action/replication_record"}},"created_at":"2026-07-05T10:21:12.906245+00:00","updated_at":"2026-07-05T10:21:12.906245+00:00"}