{"paper":{"title":"CodeT: Code Generation with Generated Tests","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"CodeT generates test cases with the same model to select correct code samples via dual execution agreement.","cross_cats":["cs.AI","cs.PL","cs.SE"],"primary_cat":"cs.CL","authors_text":"Anh Nguyen, Bei Chen, Daoguang Zan, Fengji Zhang, Jian-Guang Lou, Weizhu Chen, Zeqi Lin","submitted_at":"2022-07-21T10:18:37Z","abstract_excerpt":"The task of generating code solutions for a given programming problem can benefit from the use of pre-trained language models such as Codex, which can produce multiple diverse samples. However, a major challenge for this task is to select the most appropriate solution from the multiple samples generated by the pre-trained language models. A natural way to evaluate the quality and correctness of a code solution is to run it against a set of test cases, but the manual creation of such test cases is often costly and time-consuming. In this paper, we propose a novel method, CodeT, that leverages t"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"CodeT improves the pass@1 metric on HumanEval to 65.8%, an absolute improvement of 18.8% over the code-davinci-002 model and more than 20% over previous state-of-the-art results.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That agreement between independently generated code samples on independently generated tests reliably indicates functional correctness rather than shared bugs or test weaknesses.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"CodeT improves code generation accuracy by using the same model to create test cases and then selecting solutions via output agreement on those tests, raising HumanEval pass@1 from 47% to 65.8%.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"CodeT generates test cases with the same model to select correct code samples via dual execution agreement.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c0b39ab2ff4903c5967a93fb2485db4b52184e302c306e58cded605d5fbb8151"},"source":{"id":"2207.10397","kind":"arxiv","version":2},"verdict":{"id":"e21bdbb2-d0be-44ec-976c-a1954b6c04ea","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T23:51:28.385332Z","strongest_claim":"CodeT improves the pass@1 metric on HumanEval to 65.8%, an absolute improvement of 18.8% over the code-davinci-002 model and more than 20% over previous state-of-the-art results.","one_line_summary":"CodeT improves code generation accuracy by using the same model to create test cases and then selecting solutions via output agreement on those tests, raising HumanEval pass@1 from 47% to 65.8%.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That agreement between independently generated code samples on independently generated tests reliably indicates functional correctness rather than shared bugs or test weaknesses.","pith_extraction_headline":"CodeT generates test cases with the same model to select correct code samples via dual execution agreement."},"references":{"count":17,"sample":[{"doi":"","year":null,"title":"Program Synthesis with Large Language Models","work_id":"fd241a05-03b9-4de2-9588-9d77ce176125","ref_index":1,"cited_arxiv_id":"2108.07732","is_internal_anchor":true},{"doi":"","year":1901,"title":"Language models are few-shot learners","work_id":"f93ff324-f230-4a46-97b9-6b103c35585d","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Evaluating Large Language Models Trained on Code","work_id":"042493e9-b26f-4b4e-bbde-382072ca9b08","ref_index":3,"cited_arxiv_id":"2107.03374","is_internal_anchor":true},{"doi":"","year":null,"title":"Training Verifiers to Solve Math Word Problems","work_id":"acab1aa8-b4d6-40e0-a3ee-25341701dca2","ref_index":4,"cited_arxiv_id":"2110.14168","is_internal_anchor":true},{"doi":"","year":null,"title":"InCoder: A Generative Model for Code Infilling and Synthesis","work_id":"e98a5559-529d-48b1-833c-b85b662f190c","ref_index":5,"cited_arxiv_id":"2204.05999","is_internal_anchor":true}],"resolved_work":17,"snapshot_sha256":"0ae46256bda4d7fa4217c296ab130a7ddb6b28ef879e79fcf162e163703d5246","internal_anchors":6},"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"}