README.md

metadata

language:
  - en
pretty_name: Scientific Multitask Instructions
task_categories:
  - text-generation
  - question-answering
  - summarization
tags:
  - scientific
  - instruction-tuning
  - conversational
  - supervised-fine-tuning
  - grpo
size_categories:
  - 1K<n<10K

Scientific Multitask Instructions

A multi-task scientific instruction-following dataset created for supervised fine-tuning and preference-optimization experiments.

Dataset summary

The dataset contains 1,576 conversational scientific examples across eight task types.

Split	Examples
Train	1,260
Validation	158
Test	158
Total	1,576

Task distribution

Task	Examples
Scientific question answering	256
Summarization	220
Concept explanation	200
Method comparison	200
Technical simplification	200
Bullet generation	200
Data analysis	200
Code generation	100

Fields

• id: unique example identifier
• category: scientific subject category
• task: instruction-task category
• difficulty: example difficulty
• messages: system, user, and assistant chat messages

Usage

from datasets import load_dataset

dataset = load_dataset(
    "Miladsaeedi70/scientific-multitask-instructions",
    token=True,
)

print(dataset)
print(dataset["train"][0])

The token=True argument is needed while the repository is private.

Intended use

This dataset is intended for:

• Scientific instruction tuning
• Multi-task supervised fine-tuning
• LoRA and PEFT experiments
• Preference optimization
• GRPO reward-design research
• Scientific language-model evaluation

Split methodology

Examples were divided into train, validation, and test sets using grouped splitting to reduce prompt-template leakage between splits.

Known limitations

• The dataset is relatively small.
• Some answers may contain scientific inaccuracies.
• Task categories are not equally represented.
• Lexical-overlap metrics do not fully measure factual correctness.
• Responses should be independently verified for high-stakes use.

Publication status

The repository is initially private. Provenance, licensing, duplicate checks, and sample quality should be reviewed before public release.