roberta_chinese_base
Overview
Language model: roberta-base Model size: 392M Language: Chinese Training data: CLUECorpusSmall Eval data: CLUE dataset
Results
For results on downstream tasks like text classification, please refer to this repository.
Usage
NOTE: You have to call BertTokenizer instead of RobertaTokenizer !!!
import torch
from transformers import BertTokenizer, BertModel
tokenizer = BertTokenizer.from_pretrained("clue/roberta_chinese_base")
roberta = BertModel.from_pretrained("clue/roberta_chinese_base")
About CLUE benchmark
Organization of Language Understanding Evaluation benchmark for Chinese: tasks & datasets, baselines, pre-trained Chinese models, corpus and leaderboard.
Github: https://github.com/CLUEbenchmark Website: https://www.cluebenchmarks.com/
- Downloads last month
- 277