datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
russian-it-community-corpus
📦 Russian IT Community Corpus (RICC)
Russian IT Community Corpus (RICC) is an open, de-identified conversational dataset collected from 11 engineering community nodes spanning a 9-year timeline (2017–2026). It captures authentic discussions on backend systems, cloud infrastructure, AI/ML deployment, database internals, and software architecture.
The corpus is structured into ready-to-use splits for Instruction Fine-Tuning (SFT), Direct Preference Optimization (DPO)… See the full description on the dataset page: https://huggingface.co/datasets/wwewtech/russian-it-community-corpus.russian-nmo-medical-mcq
Russian NMO Medical MCQ
Choose language / Выберите язык: Русский | English
Русский
Это датасет русскоязычных медицинских тестовых вопросов НМО с вариантами ответа.
В нем есть вопросы с одним правильным вариантом и вопросы с несколькими правильными
вариантами. Датасет подготовлен так, чтобы его можно было сразу использовать для
тонкой настройки LLM, проверки качества ответов и экспериментов с медицинским QA.
Главная идея простая: дать модели вопрос, тему и варианты ответа… See the full description on the dataset page: https://huggingface.co/datasets/drkolesnikov/russian-nmo-medical-mcq.MMLU-Pro_Kazakh_Russian
Dataset Summary
These are the machine-translated Kazakh and Russian versions of the MMLU-Pro (Massive Multitask Language Understanding Pro) dataset (test set).
These datasets are used to test the world knowledge and problem-solving capabilities of large language models across a vast range of subjects in the Kazakh and Russian languages. As an enhanced version of the original MMLU, it serves as a more rigorous benchmark for evaluating how well models understand complex academic… See the full description on the dataset page: https://huggingface.co/datasets/issai/MMLU-Pro_Kazakh_Russian.code-alchemy-rust
CodeAlchemy Rust
Rust-only derivative of open-alchemy/code-alchemy. It preserves the five training configs, two evaluation configs, original splits, row order, columns, values, and task/evaluation fields.
Rows were selected from the source-native language labels:
Rust and rust in training data and dev-eval
rs in trace-eval
Labels remain unchanged in the output. code-trace.external_packages is normalized to list<string> because source Parquet shards physically alternate between… See the full description on the dataset page: https://huggingface.co/datasets/adityabhushannagar/code-alchemy-rust.rust-forum-qa-pairs
Rust Programming QA Pairs
Dataset Description
The Rust Programming QA Pairs dataset is a collection of question-answer pairs extracted from the Rust programming language user forums. It contains high-quality programming questions and their accepted answers, focusing on Rust programming language topics. The dataset is designed to support natural language processing tasks related to programming assistance, code understanding, and technical question answering.
Each entry… See the full description on the dataset page: https://huggingface.co/datasets/portex/rust-forum-qa-pairs.RusFinQABenchmark
RusFinQABenchmark
Результаты оценки шести больших языковых моделей на датасете RuFinQA с использованием системы метрик FinCoT-Eval.
Модели
gemma2:9b
qwen2.5:7b
deepseek-r1:7b
phi3:3.8b
llama3.1:8b
aya:8b
Метрики
FinCoT-Eval: FAA, OT, CSPS, NEPS, FCS, Composite
Текстовые: COMET, BERTScore, ROUGE, BLEU
Структура файлов
evaluation.csv — построчные метрики для 6000 генераций (1000 вопросов × 6 моделей)
summary.csv — агрегированные… See the full description on the dataset page: https://huggingface.co/datasets/RusNLPWorld/RusFinQABenchmark.ru-stem-dialogues
Russian STEM Educational Dialogues
Описание
Синтетический датасет русскоязычных учебных диалогов по STEM-темам (математика, физика, химия,
биология, информатика, программирование, инженерия). Каждый диалог — реалистичное взаимодействие
между пользователем (школьник / студент / профессионал) и ассистентом.
Методология
Модель: Qwen/Qwen2.5-7B-Instruct (4-bit NF4 quantization, bitsandbytes)
Формат генерации: текстовый формат с разделителями… See the full description on the dataset page: https://huggingface.co/datasets/AtesiT/ru-stem-dialogues.RuFinQA
🏦 RusFinChain
RusFinChain is a Russian benchmark for evaluating Large Language Models (LLMs) on financial analysis tasks with Ground-Truth Chain-of-Thought.
📊 Overview
Total questions: 44,627
Task types: 7
Skills: 12
Difficulty levels: 3
Version: 3.3.0
Language: Russian
📖 Source Material & Data Licensing
Foundational SourceThe methodological framework, financial formulas, problem typology, and a significant portion of the practical… See the full description on the dataset page: https://huggingface.co/datasets/RusNLPWorld/RuFinQA.RusFinChain
RusFinChain — Symbolic Financial Reasoning in Russian
RusFinChain is a large-scale symbolic financial reasoning benchmark in Russian, designed to evaluate the multi‑step reasoning capabilities of large language models (LLMs). It contains 5,280 tasks across 17 domains and 172 topics, with three difficulty levels: Basic (Базовый), Intermediate (Средний), Advanced (Продвинутый).
Each task includes a natural‑language question, a step‑by‑step solution, a final numeric answer, a LaTeX… See the full description on the dataset page: https://huggingface.co/datasets/RusNLPWorld/RusFinChain.kz-rus-articles-comprehensive
🇰🇿🇷🇺 Kazakh-Russian Articles Comprehensive Dataset
A high-quality bilingual corpus for cross-lingual NLP research
📋 Dataset Overview
The Kazakh-Russian Articles Comprehensive Dataset is a meticulously curated bilingual corpus designed to advance natural language processing research for Kazakh and Russian languages. This dataset addresses the critical need for high-quality parallel and comparable text resources in Central Asian language pairs, particularly… See the full description on the dataset page: https://huggingface.co/datasets/Adilbai/kz-rus-articles-comprehensive.aibe-xxi-set-a
AIBE 16–21 MCQ Benchmark
All 100 multiple-choice questions from each of the last six All India Bar Examinations (AIBE XVI–XXI, 2021–2026), as six configs of one dataset, each paired with the most authoritative answer key available and labeled with the same official 19-subject BCI syllabus taxonomy.
Config
Exam
Held
Set
Question source
Answer key
Withdrawn
Multi-answer
aibe16
AIBE XVI
Oct 2021
C
Delhi Law Academy compilation
DLA 4-set key table (no official copy… See the full description on the dataset page: https://huggingface.co/datasets/rushankg/aibe-xxi-set-a.RuFinQA-lite
RuFinQA — A Massive Multi-Task Reasoning Benchmark for Russian Financial Report Understanding
RuFinQA is a large-scale multi-task benchmark designed to evaluate the ability of language models to understand and reason over Russian statutory financial reports (Balance Sheet, Income Statement, Cash Flow Statement).
It contains 36,330 question–answer pairs across 5 task types, automatically derived from real-world corporate accounting statements obtained from open government data… See the full description on the dataset page: https://huggingface.co/datasets/RusNLPWorld/RuFinQA-lite.