mechtaev

4 ms·

1.
Reducing Hallucinations in LLM-Gen. Code via Semantic Triangulation (OOPSLA 26)(github.com)
2 points by mechtaev 2mo ago|0 comments
2.
Automated Repair of Ambiguous Problem Descriptions for LLM-Based Code Generation(github.com)
2 points by mechtaev 1y ago|0 comments
3.
HoarePrompt: Structural Reasoning About Program Correctness in Natural Language(github.com)
6 points by mechtaev 2y ago|0 comments
4.
AutoCodeRover: Autonomous Program Improvement(github.com)
96 points by mechtaev 2y ago|60 comments
5.
by mechtaev 2y ago
https://arxiv.org/abs/2404.05427
6.
by mechtaev 4y ago
Souffle Datalog is used for defining program analyses in such projects like Doop [1] for Java and cclyzer++ [2] for LLVM. GitHub's CodeQL [3] is another Datalog dialect used for detecting bugs and vulnerabilities. Datomic [4] is a data
7.
by mechtaev 4y ago
Disclaimer: I am a co-author of Modus. Modus is not really an alternative to Nix as a package manager, but to Nix as a language. It has several advantages, for example, it is non-Turing-complete, so there can not be infinite loops, and its
8.
by mechtaev 4y ago
Apart from caching, BuildKit also provides automatic parallelization. Because of these two features, we use BuildKit to make our research results, experiments with program analysis tools on software benchmarks, easier to share and faster to
9.
by mechtaev 4y ago
Hey HN, we are happy to announce Modus, a language for building Docker/OCI container images. Modus uses logic programming to express interactions among build parameters, specify complex build workflows, automatically parallelise and ca
10.
Show HN: Modus, a Datalog dialect for building Docker/OCI container images(modus-continens.com)
9 points by mechtaev 4y ago|1 comments
11.
by mechtaev 5y ago
fzf can partially solve this, as shown in this blog post: https://nickjanetakis.com/blog/fuzzy-search-your-bash-histor...
12.
by mechtaev 5y ago
SyGuS competition (SyGuS-Comp)[1] is a widely-used collection of such problems, but of course there will always be problem domains that are not well-represented by standard benchmarks. 1. https://sygus.org/