Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
parquet
Sub-tasks:
sentiment-scoring
Size:
1K - 10K
{"splits": ["train", "test"]} |