Datasets:
Tasks:
Text2Text Generation
Modalities:
Text
Formats:
parquet
Languages:
English
Size:
10K - 100K
License:
metadata
annotations_creators:
- machine-generated
language:
- en
language_creators:
- expert-generated
license:
- cc-by-4.0
multilinguality:
- monolingual
pretty_name: AMR 3.0 Parsed
size_categories:
- 10K<n<100K
source_datasets:
- ldc2020t02
task_categories:
- text2text-generation
Dataset Card for AMR 3.0 Parsed
Dataset Description
- Repository: LDC2020T02
- Paper: AMR Paper
- Point of Contact: [Your Contact Info]
Dataset Summary
This dataset contains parsed Abstract Meaning Representation (AMR) annotations from the LDC2020T02 release, formatted as instruction-following conversations. Each example consists of a sentence and its corresponding AMR graph representation.
Supported Tasks and Leaderboards
- Tasks: Semantic parsing, specifically generating AMR graphs from English sentences
- Leaderboards: AMR Parsing
Languages
The dataset is in English.
Dataset Structure
Data Instances
An example looks like:
{
'conversations': [
{
'role': 'user',
'content': 'Generate an Abstract Meaning Representation (AMR) graph for the following sentence: [sentence]'
},
{
'role': 'assistant',
'content': '[AMR graph]'
}
]
}