Removing Non-Stationary Knowledge From Pre-Trained Language Models for Entity-Level Sentiment Classification in Finance
Paper • 2301.03136 • Published
KorFinASC-XLM-RoBERTa
Pretrained XLM-RoBERTA-Large transfered to the Finance domain on Korean Language.
See paper for more details.
Data
KorFinASC-XLM-RoBERTa is extensively trained on multiple datasets including KorFin-ASC, Ko-FinSA, Ko-ABSA and ModuABSA.
How to use.
>>> from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
>>> tokenizer = AutoTokenizer.from_pretrained("amphora/KorFinASC-XLM-RoBERTa")
>>> model = AutoModelForSequenceClassification.from_pretrained("amphora/KorFinASC-XLM-RoBERTa")
>>> input_str = "장 전체가 폭락한 가운데 삼성전자만 상승세를 이어갔다. </s> 삼성전자"
>>> input = tokenizer(input_str, return_tensors='pt')
>>> output = model.generate(**input, max_length=20)
- Downloads last month
- 197