Kotlin ML Pack: Technical Report

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Titov, Sergey, Evtikhiev, Mikhail, Shapkin, Anton, Smirnov, Oleg, Boytsov, Sergei, Karaeva, Dariia, Sheptyakov, Maksim, Arkhipov, Mikhail, Bryksin, Timofey, Bogomolov, Egor
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916265231122432
author Titov, Sergey
Evtikhiev, Mikhail
Shapkin, Anton
Smirnov, Oleg
Boytsov, Sergei
Boytsov, Sergei
Karaeva, Dariia
Sheptyakov, Maksim
Arkhipov, Mikhail
Bryksin, Timofey
Bogomolov, Egor
author_facet Titov, Sergey
Evtikhiev, Mikhail
Shapkin, Anton
Smirnov, Oleg
Boytsov, Sergei
Boytsov, Sergei
Karaeva, Dariia
Sheptyakov, Maksim
Arkhipov, Mikhail
Bryksin, Timofey
Bogomolov, Egor
contents In this technical report, we present three novel datasets of Kotlin code: KStack, KStack-clean, and KExercises. We also describe the results of fine-tuning CodeLlama and DeepSeek models on this data. Additionally, we present a version of the HumanEval benchmark rewritten by human experts into Kotlin - both the solutions and the tests. Our results demonstrate that small, high-quality datasets (KStack-clean and KExercises) can significantly improve model performance on code generation tasks, achieving up to a 16-point increase in pass rate on the HumanEval benchmark. Lastly, we discuss potential future work in the field of improving language modeling for Kotlin, including the use of static analysis tools in the learning process and the introduction of more intricate and realistic benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2405_19250
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Kotlin ML Pack: Technical Report
Titov, Sergey
Evtikhiev, Mikhail
Shapkin, Anton
Smirnov, Oleg
Boytsov, Sergei
Boytsov, Sergei
Karaeva, Dariia
Sheptyakov, Maksim
Arkhipov, Mikhail
Bryksin, Timofey
Bogomolov, Egor
Software Engineering
Artificial Intelligence
Programming Languages
In this technical report, we present three novel datasets of Kotlin code: KStack, KStack-clean, and KExercises. We also describe the results of fine-tuning CodeLlama and DeepSeek models on this data. Additionally, we present a version of the HumanEval benchmark rewritten by human experts into Kotlin - both the solutions and the tests. Our results demonstrate that small, high-quality datasets (KStack-clean and KExercises) can significantly improve model performance on code generation tasks, achieving up to a 16-point increase in pass rate on the HumanEval benchmark. Lastly, we discuss potential future work in the field of improving language modeling for Kotlin, including the use of static analysis tools in the learning process and the introduction of more intricate and realistic benchmarks.
title Kotlin ML Pack: Technical Report
topic Software Engineering
Artificial Intelligence
Programming Languages
url https://arxiv.org/abs/2405.19250