Datasets:
annotations_creators: []
language:
- en
language_creators: []
license: []
multilinguality:
- monolingual
pretty_name: adaption-no_robots_instructions
size_categories:
- 10K<n<100K
source_datasets:
- extended|https://huggingface.co/datasets/HuggingFaceH4/no_robots
tags:
- adaption
- instruction-tuning
- writing-editing-communication
- entertainment
- history
task_categories: []
task_ids: []
This dataset is a remastered version of this dataset prepared using Adaption's Adaptive Data platform.
adaption-no_robots_instructions
This dataset contains 10,000 human-written instruction-response pairs covering diverse topics such as creative writing, factual queries, and practical advice. Unlike many contemporary datasets, it was explicitly curated without using AI-generated content to ensure authentic human phrasing and reasoning. The data is formatted as conversation lists with user prompts and assistant responses, suitable for training and evaluating language models on genuine human interaction patterns.
Dataset size
There are 20,000 data points in this dataset. This is an instruction tuning dataset.
Quality of Remastered Dataset
The final quality is B, with a relative quality improvement of 7.1%.
Domain
- Writing-editing-communication (28%)
- Entertainment (8%)
- History (8%)
Language
- English (100%)
Tone
- Informative (36%)
- Creative (8%)
- Engaging (6%)
Evaluation Results
Quality Gains:
Grade Improvement:
Percentile Chart:
