Datasets:
Tasks:
Text Classification
Sub-tasks:
multi-class-classification
Size:
100K<n<1M
ArXiv:
Tags:
relation extraction
License:
added a todo
Browse files
README.md
CHANGED
@@ -67,6 +67,12 @@ task_ids:
|
|
67 |
- **Size of the generated dataset:** 40.9 MB
|
68 |
- **Total amount of disk used:** 103.2 MB
|
69 |
### Dataset Summary
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
NOTE: This Datasetreader supports a reduced version of the original TACRED JSON format with the following changes:
|
71 |
- Removed fields: stanford_pos, stanford_ner, stanford_head, stanford_deprel, docid
|
72 |
|
|
|
67 |
- **Size of the generated dataset:** 40.9 MB
|
68 |
- **Total amount of disk used:** 103.2 MB
|
69 |
### Dataset Summary
|
70 |
+
|
71 |
+
TODO initial dataset reader for MultiTACRED, needs to be tested / finalized with dataset. Should use the the TACRED
|
72 |
+
dataset reader (https://huggingface.co/datasets/DFKI-SLT/tacred) code but supply other download URLs per language,
|
73 |
+
with language = config item. e.g. 'en-original', 'de-revised', 'ar-retacred' etc.
|
74 |
+
|
75 |
+
|
76 |
NOTE: This Datasetreader supports a reduced version of the original TACRED JSON format with the following changes:
|
77 |
- Removed fields: stanford_pos, stanford_ner, stanford_head, stanford_deprel, docid
|
78 |
|