Atmanstr/MultiDomain_Instruction

---
license: apache-2.0
task_categories:
- text-classification
- question-answering
- summarization
language:
- en
- ru
- es
- fr
tags:
- math
- science
- reasoning
- web
---
# MultiDomain_Instruction

A multi-domain instruction dataset designed for instruction tuning and supervised fine-tuning (SFT) of large language models. The dataset contains tasks from multiple domains such as question answering, summarization, reasoning, classification, and general knowledge to improve model generalization.

Overview

MultiDomain_Instruction is created to support instruction-following training for LLMs. Instead of focusing on a single task, this dataset combines multiple domains and task types to help models learn broader reasoning and response capabilities.

---This dataset is a merged instruction-tuning dataset for training large language models (LLMs). It combines multiple high-quality sources, including:

HuggingFaceH4/CodeAlpaca_20K – programming tasks and code explanations

meta-math/MetaMathQA – math word problems

databricks/databricks-dolly-15k – instruction-following tasks

OASST – human-assistant conversation pairs

lhpku20010120/WebInstruct_Full– web instruction tasks

KonstantyM/science_qa – science questions and answers

Alibaba-Apsara/Superior-Reasoning-SFT-gpt-oss-120b - Reasoning Abilities

---The dataset is designed for instruction tuning and instruction-following LLMs.

---Dataset Structure

Split:

train – training examples

validation – evaluation examples

Columns:

Column Description
instruction The instruction or question to follow
input Optional input or context (can be empty)
output The expected response or solution

Example:

{
"instruction": "Calculate the area of a circle with radius 5.",
"input": "",
"output": "To calculate the area of a circle, use the formula πr². The area is 3.1416 * 5² = 78.54."
}

---
author:Atmanstr
year:2026
---