klue-roberta-base-kornli

This model trained with Korean dataset. Input premise sentence and hypothesis sentence. You can use English, but don't expect accuracy. If the context is longer than 1200 characters, the context may be cut in the middle and the result may not come out well.

KLUE-RoBERTa-base-KorNLI DEMO: Ainize DEMO

KLUE-RoBERTa-base-KorNLI API: Ainize API

Overview

Language model: klue/roberta-base

Language: Korean

Training data: kakaobrain KorNLI

Eval data: kakaobrain KorNLI

Code: See Ainize Workspace

Usage

In Transformers

from transformers import AutoTokenizer, pipeline

tokenizer = AutoTokenizer.from_pretrained("ehdwns1516/klue-roberta-base-kornli")

classifier = pipeline(
    "text-classification",
    model="ehdwns1516/klue-roberta-base-kornli",
    return_all_scores=True,
)

premise = "your premise"
hypothesis = "your hypothesis"

result = dict()
result[0] = classifier(premise + tokenizer.sep_token + hypothesis)[0]