语言:

• tr 标签:
• roberta 许可证: cc-by-nc-sa-4.0 数据集:
• oscar

RoBERTa 土耳其语中型WordPiece 44k（无大小写区分）

该模型基于土耳其语，采用掩码语言建模（MLM）目标进行预训练。模型不区分大小写。 预训练语料来自OSCAR的土耳其语部分，但经过进一步过滤和清洗。

模型架构与bert-medium类似（8层，8个头，隐藏层大小为512）。分词算法采用WordPiece。词汇表大小为44.5k。

详细信息可参阅以下论文： https://arxiv.org/abs/2204.08832

以下代码可用于加载模型和分词，示例最大长度（514）可调整：

	model = AutoModel.from_pretrained([model_path])
	#用于序列分类：
	#model = AutoModelForSequenceClassification.from_pretrained([model_path], num_labels=[num_classes])

	tokenizer = PreTrainedTokenizerFast(tokenizer_file=[file_path])
	tokenizer.mask_token = "[MASK]"
	tokenizer.cls_token = "[CLS]"
	tokenizer.sep_token = "[SEP]"
	tokenizer.pad_token = "[PAD]"
	tokenizer.unk_token = "[UNK]"
	tokenizer.bos_token = "[CLS]"
	tokenizer.eos_token = "[SEP]"
	tokenizer.model_max_length = 514

BibTeX条目及引用信息

@misc{https://doi.org/10.48550/arxiv.2204.08832,
  doi = {10.48550/ARXIV.2204.08832},
  url = {https://arxiv.org/abs/2204.08832},
  author = {Toraman, Cagri and Yilmaz, Eyup Halit and Şahinuç, Furkan and Ozcelik, Oguzhan},
  keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {Impact of Tokenization on Language Models: An Analysis for Turkish},
  publisher = {arXiv},
  year = {2022},
  copyright = {Creative Commons Attribution Non Commercial Share Alike 4.0 International}
}