README.md

metadata

language:
  - ja
library_name: transformers
license: apache-2.0
metrics:
  - accuracy
  - f1
pipeline_tag: text-classification
datasets:
  - LoneWolfgang/japanese-twitter-sentiment

BERT for Sentiment Analysis of Japanese Twitter

This model was finetuned from BERT for Japanese Twitter, which was adapted from Japanese BERT by Tohoku NLP by continuing MLM on a Twitter corpus.

It used Japanese Twitter Sentiment 1k (JTS1k) for finetuning, omitting the mixed examples.

Labels

0 -> Negative; 1 -> Neutral; 2 -> Positive

Example Pipeline

from transformers import pipeline
sentiment = pipeline("sentiment-analysis", model="LoneWolfgang/bert-for-japanese-twitter-sentiment")
sentiment ("こちらのカフェ、サービスが残念でした。二度と行かないかな…😞 #がっかり")

[{'label': 'negative', 'score': 0.8242}]