pretty_name: Detailed vs Vague Pairs (english)
language:
- en
license: cc-by-4.0
size_categories: - n<1K
task_categories: - text-generation
- text-classification
tags: - abliteration
- text
- synthetic
- contrastive
- specificity
- vagueness
- ambiguity
- implicit-detail
- datasets
configs: - config_name: explicit_specificity
default: true
data_files:- split: train
path: data/explicit_specificity/train.parquet
- split: train
- config_name: implicit_richness
data_files:- split: train
path: data/implicit_richness/train.parquet
- split: train
Detailed vs Vague Pairs (english)
A sentence can spell out the details. It can also describe a rich experience or a deep understanding while leaving those details unnamed.
This collection keeps those two ideas separate. It contains 700 English text pairs for exploring explicit specificity and implicit richness, with a dedicated subset for each.
Explore: Explicit specificity · Implicit richness · Columns · Load the data
General properties
| Property | Value |
|---|---|
| Version | 1.0.0 |
| Pairs | 700 |
| Subsets | 600 explicit specificity · 100 implicit richness |
| Language | English |
| Coverage | 145 domains |
| Formats | Parquet for the viewer · JSONL mirrors |
| Data size | 147 kB across both Parquet files |
| License | CC BY 4.0 |
Explicit specificity
The 600 pairs compare a detailed formulation with a concise, vague, or ambiguous counterpart. Some compress the wording. Others leave information unstated or allow more than one reading. A compact counterpart can still be precise.
The subset covers 50 styles in vague_style with 12 pairs each and 40 detailed styles with 15 pairs each. Style labels identify the dominant trait of an example. Closely related labels can overlap.
Implicit richness
The 100 pairs express richer or less defined qualities without listing the underlying particulars. The scene stays the same, while the perception, understanding, memory, portrayal, or object quality changes.
These pairs may describe opposing qualities or request different levels of richness. The two formulations are comparable in length, so a richer version does not depend on adding a longer list of details.
Load the data
Choose the subset that matches the distinction you want to study.
from datasets import load_dataset
explicit = load_dataset(
"strangecreator/detailed-vs-vague-pairs-english",
"explicit_specificity",
split="train",
)
implicit = load_dataset(
"strangecreator/detailed-vs-vague-pairs-english",
"implicit_richness",
split="train",
)
Each subset has one train split containing all its pairs. Create a task-specific evaluation partition when needed, keeping the two texts from each row together.
Load the files locally
from datasets import load_dataset
explicit = load_dataset(
"parquet",
data_files="data/explicit_specificity/train.parquet",
split="train",
)
implicit = load_dataset(
"parquet",
data_files="data/implicit_richness/train.parquet",
split="train",
)
Columns
Both subsets use detailed_text and vague_text as their first two columns. In the explicit subset, vague_text can be concise, vague, or ambiguous. In the implicit subset, it expresses the less rich or less defined quality. The remaining columns describe the domain, register, and intended contrast. The full schema is also available as JSON.
Column reference
Shared columns
| Column | Meaning |
|---|---|
domain |
A broad subject area from the original taxonomy. |
register |
The assigned register: conversational, neutral, or formal. |
contrast_explanation |
A short explanation of the intended difference between the two formulations. |
Explicit specificity columns
| Column | Meaning |
|---|---|
detailed_text |
The detailed formulation in an explicit specificity pair. |
vague_text |
Its concise, vague, or ambiguous counterpart. The style label specifies the intended change. |
detailed_style |
The dominant style of the detailed formulation, such as Precise, Granular, or Context-rich. |
vague_style |
The dominant style of vague_text, such as Concise, Vague, or Scope-ambiguous. |
Implicit richness columns
| Column | Meaning |
|---|---|
detailed_text |
The formulation expressing greater qualitative richness, clarity, or distinctiveness. |
vague_text |
The formulation expressing less richness, clarity, or distinctiveness. |
contrast_dimension |
The primary implicit dimension: perception, understanding, memory, representation, or intrinsic_features. |
Creation and review
The examples were authored with AI assistance (Codex: GPT-6.1-Sol-Extra) using a planned taxonomy. Each pair received an author review and a separate review by another AI subagent. The checks covered the intended contrast, shared context, naturalness, and annotation consistency.
Coverage summary is included.
Scope and limitations
This is a small, deliberately balanced synthetic collection for controlled experiments with LLMs and agentic systems, including exploratory abliteration studies.
Semantic review was performed by AI agents. The annotations support grouping and comparing examples. Boundaries may overlap between some styles.
Credits and license
Created by strangecreator. The dataset is available under CC BY 4.0, allowing reuse and adaptation with attribution. See LICENSE for the full terms.
Citation
@misc{strangecreator2026englishdetailedvsvaguepairs,
author = {strangecreator},
title = {Detailed vs Vague Pairs (english)},
year = {2026},
howpublished = {Hugging Face dataset},
url = {https://huggingface.co/datasets/strangecreator/detailed-vs-vague-pairs-english}
}