Dataset · ibm-research
English argument quality and stance dataset
ibm-research/argument_quality_ranking_30k
Over 30,000 English arguments on controversial topics with quality and stance labels for training classifiers.
What's inside
Contains 30,497 crowdsourced English text arguments across 71 debatable topics. Each example is annotated for argument quality and author stance, with pre-made train, validation, and test splits. Published by IBM researchers alongside a research paper, it also includes the 9,487-record Argument Topic subset.
What you can do with it
- Training models to rank and evaluate text persuasiveness
- Classifying author stance (pro/con) in discussions and debates
- Fine-tuning LLMs to analyze argumentation and identify weak points in text
- Tasks
- Text classification
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 1.6K
- Likes
- 17
Models trained on it
- mradermacher/elf-1-12b-GGUF 737
- mradermacher/elf-1-31b-GGUF 710
- mradermacher/bobcat-1.3-GGUF 263
- sanghwa-na/bobcat-1.3Zero shot classification 232
- LatentAILabs/yama-indicZero shot classification 173
- LatentAILabs/laya-indicText classification 127
- arcbright/elf-1-12bText classification 104
- arcbright/elf-1-31bText classification 63
How to load it
from datasets import load_dataset
ds = load_dataset("ibm-research/argument_quality_ranking_30k")