References:
Use the following command to load this dataset in TFDS:
ds = tfds.load('huggingface:limit')
- Description:
Motion recognition is one of the basic cognitive capabilities of many life forms, yet identifying motion of physical entities in natural language have not been explored extensively and empirically. Literal-Motion-in-Text (LiMiT) dataset, is a large human-annotated collection of English text sentences describing physical occurrence of motion, with annotated physical entities in motion.
- License: No known license
- Version: 1.0.0
- Splits:
| Split | Examples |
|---|---|
'test' |
1000 |
'train' |
23559 |
- Features:
{
"id": {
"dtype": "int32",
"id": null,
"_type": "Value"
},
"sentence": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"motion": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"motion_entities": [
{
"entity": {
"dtype": "string",
"id": null,
"_type": "Value"
},
"start_index": {
"dtype": "int32",
"id": null,
"_type": "Value"
}
}
]
}