Semantic scaffolds for pseudocode-to-code generation
… to the SPoC dataset for pseudocodeto-code generation, in … pseudocode annotations and
aim to produce a program satisfy… Search-based Pseudocode to Code (SPoC) dataset (Kulal et …
aim to produce a program satisfy… Search-based Pseudocode to Code (SPoC) dataset (Kulal et …
C3PO: A lightweight copying mechanism for translating pseudocode to code
… We divide the task of pseudocode to code translation into three … C3PO has been trained on
the SPoC dataset (Kulal et al.… In the SPoC dataset, we observed that the mean ratio of copied …
the SPoC dataset (Kulal et al.… In the SPoC dataset, we observed that the mean ratio of copied …
Knowledge Transfer for Pseudo-code Generation from Low Resource Programming Language
… On the similar lines of Kulal et al. [2019] we use Best-first-search to combine code lines …
We consider the SPoC original test splits TESTP and TESTW as opposed to the random splits …
We consider the SPoC original test splits TESTP and TESTW as opposed to the random splits …
Leveraging Statistical Machine Translation for Code Search
H Phan, A Jannesari - Proceedings of the 28th International Conference …, 2024 - dl.acm.org
… code compared to expected code, including metrics like CodeBLEU [31] and ParaBLEU [44].
With advancements in … In a manner similar to pseudocode-to-code training using Machine …
With advancements in … In a manner similar to pseudocode-to-code training using Machine …
Large language models as analogical reasoners
… In line with existing work on code generation (Li et al., 2022b; Chen et al., 2023), we report
the Acc@1 and Acc@10 metrics. Acc@k measures whether at least one of the k sampled …
the Acc@1 and Acc@10 metrics. Acc@k measures whether at least one of the k sampled …
Communicating natural programs to humans and machines
… systems that are capable of solving general, procedural tasks such as ARC? Traditional
approaches of program … – for any task, there exists a program, written in a predefined Domain …
approaches of program … – for any task, there exists a program, written in a predefined Domain …
Compilable neural code generation with compiler feedback
… code using an energy model with compilability constraints. This paper focuses on the
task of compilable neural code … In Advances in Neural Information Processing Systems 32: …
task of compilable neural code … In Advances in Neural Information Processing Systems 32: …
Test-Driven Multi-Task Learning with Functionally Equivalent Code Transformation for Neural Code Generation
… Despite much progress on code generation, we argue that these … Yasunaga and Liang [35]
designed a program-feedback … In Advances in Neural Information Processing Systems 32: …
designed a program-feedback … In Advances in Neural Information Processing Systems 32: …
Boldly Going Where No Benchmark Has Gone Before: Exposing Bias and Shortcomings in Code Generation Evaluation
… in the popularity of large language models (LLMs) has also been accompanied by
significant advancements in code-generation LLMs (Code-LLMs). Prominent examples of such …
significant advancements in code-generation LLMs (Code-LLMs). Prominent examples of such …
Re (gEx| DoS) Eval: Evaluating Generated Regular Expressions and their Proneness to DoS Attacks
… With the recent advances of code generation techniques based on Large Language
Models (LLMs), developers are … Advances in neural information processing systems 31 (2018). …
Models (LLMs), developers are … Advances in neural information processing systems 31 (2018). …