{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:L5Y7CXQ2SEJPGXTB4BB6CAPGVK","short_pith_number":"pith:L5Y7CXQ2","schema_version":"1.0","canonical_sha256":"5f71f15e1a9112f35e61e043e101e6aa806d551491c0122018aaaa98597dafdd","source":{"kind":"arxiv","id":"2305.13514","version":2},"attestation_state":"computed","paper":{"title":"Small Language Models Improve Giants by Rewriting Their Outputs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aliaksei Severyn, Arthur Bra\\v{z}inskas, Eric Malmi, Giorgos Vernikos, Jakub Adamek, Jonathan Mallinson","submitted_at":"2023-05-22T22:07:50Z","abstract_excerpt":"Despite the impressive performance of large language models (LLMs), they often lag behind specialized models in various tasks. LLMs only use a fraction of the existing training data for in-context learning, while task-specific models harness the full dataset for fine-tuning. In this work, we tackle the problem of leveraging training data to improve the performance of LLMs without fine-tuning. Our approach directly targets LLM predictions without requiring access to their weights. We create a pool of candidates from the LLM through few-shot prompting and we employ a compact model, the LM-correc"},"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":"2305.13514","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-05-22T22:07:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"3f538f03d041befbc4c6490cf6c6e49cfda8a13b002e6280ce7712b3fe22b7bb","abstract_canon_sha256":"4cc5e19c9754311bc5527c83a26213446319da9ef29e6a3c9366005c35dae07e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:39:58.865885Z","signature_b64":"V6jqKjiph+DcwtKv0dXKKHQVyzPlyy279javnxiyVRWvnzL9gsYXbA+vppfTA7kaQk5PWqxfS66ePROlZtCDDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5f71f15e1a9112f35e61e043e101e6aa806d551491c0122018aaaa98597dafdd","last_reissued_at":"2026-07-05T07:39:58.865506Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:39:58.865506Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Small Language Models Improve Giants by Rewriting Their Outputs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aliaksei Severyn, Arthur Bra\\v{z}inskas, Eric Malmi, Giorgos Vernikos, Jakub Adamek, Jonathan Mallinson","submitted_at":"2023-05-22T22:07:50Z","abstract_excerpt":"Despite the impressive performance of large language models (LLMs), they often lag behind specialized models in various tasks. LLMs only use a fraction of the existing training data for in-context learning, while task-specific models harness the full dataset for fine-tuning. In this work, we tackle the problem of leveraging training data to improve the performance of LLMs without fine-tuning. Our approach directly targets LLM predictions without requiring access to their weights. We create a pool of candidates from the LLM through few-shot prompting and we employ a compact model, the LM-correc"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.13514","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/2305.13514/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":"2305.13514","created_at":"2026-07-05T07:39:58.865560+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.13514v2","created_at":"2026-07-05T07:39:58.865560+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.13514","created_at":"2026-07-05T07:39:58.865560+00:00"},{"alias_kind":"pith_short_12","alias_value":"L5Y7CXQ2SEJP","created_at":"2026-07-05T07:39:58.865560+00:00"},{"alias_kind":"pith_short_16","alias_value":"L5Y7CXQ2SEJPGXTB","created_at":"2026-07-05T07:39:58.865560+00:00"},{"alias_kind":"pith_short_8","alias_value":"L5Y7CXQ2","created_at":"2026-07-05T07:39:58.865560+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2501.01668","citing_title":"CoT-based Synthesizer: Enhancing LLM Performance through Answer Synthesis","ref_index":34,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK","json":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK.json","graph_json":"https://pith.science/api/pith-number/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/graph.json","events_json":"https://pith.science/api/pith-number/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/events.json","paper":"https://pith.science/paper/L5Y7CXQ2"},"agent_actions":{"view_html":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK","download_json":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK.json","view_paper":"https://pith.science/paper/L5Y7CXQ2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.13514&json=true","fetch_graph":"https://pith.science/api/pith-number/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/graph.json","fetch_events":"https://pith.science/api/pith-number/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/action/storage_attestation","attest_author":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/action/author_attestation","sign_citation":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/action/citation_signature","submit_replication":"https://pith.science/pith/L5Y7CXQ2SEJPGXTB4BB6CAPGVK/action/replication_record"}},"created_at":"2026-07-05T07:39:58.865560+00:00","updated_at":"2026-07-05T07:39:58.865560+00:00"}