The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.

Multilingual Code and Language Dataset (Central Asian & Global)

This repository contains a curated dataset designed for training, fine-tuning, and evaluating multi-language AI models, specifically focusing on cross-lingual code understanding, documentation synthesis, and natural language processing (NLP) for Central Asian languages.

Dataset Summary

The dataset bridges the gap between high-resource and low-resource languages in the context of software engineering and technical documentation. It combines source code from multiple programming paradigms with natural language commentary, technical texts, and translations.

Key Metrics:

Programming Languages: Python, Rust, C++, Go
Human Languages: Kyrgyz (KY) 23 GB, Russian (RU) 60 GB, English (EN) 1.5 GB, Tajik (TG) 7 GB, Kazakh (KK) 22 GB, Uzbek (UZ) 7 GB
Total file size after compression: ~30 GB .7z file
Total size 150 GB

Dataset Structure

The repository is organized by language domains and technical tasks to facilitate standard machine learning training pipelines (train / validation splits):

/python — Python source code, docstrings, and algorithmic tasks.
/rust — Rust crates, syntax patterns, and memory-safety documentation.
/cpp — Legacy and modern C++ codebases with multi-language comments.
/go — Go microservices, concurrent design patterns, and structural data.
/docs — Parallel technical corpora, translations, and linguistic evaluation benchmarks in KY, RU, EN, TG, KK, and UZ.

Use Cases & Applications

Code LLM Fine-Tuning: Enhancing models like StarCoder or CodeLlama to better understand developer comments in Central Asian languages.
Technical Translation: Training machine translation models on highly specialized technical and programming terminology.
Cross-Lingual Information Retrieval: Developing search engines capable of mapping queries from regional languages to multi-language source code bases.

Licensing & Attribution

This dataset is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. The data consists of open-source components, public domain texts, and synthetic benchmarks generated for research purposes.

Downloads last month: 41

Total file size:

27.7 GB