metadata
dataset_info:
features:
- name: sentence1
dtype: string
- name: sentence2
dtype: string
- name: label
dtype: int64
splits:
- name: train
num_bytes: 456693.60931899643
num_examples: 669
- name: test
num_bytes: 23210.138590203107
num_examples: 34
- name: valid
num_bytes: 91475.25209080048
num_examples: 134
download_size: 289139
dataset_size: 571379
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: valid
path: data/valid-*
language:
- en
- ne
Paraphrase data set containing paraphrases in either English or Nepali. Suitable to train multi-lingual similarity model. The dataset was generated using several LLMs including Gemini 2.5 Flash and GPT 4.1 mini.
The label column indicates whether the pair are paraphrases of each other or not. Currently about 81% of pairs have label=1 i.e. paraphrases.