ctoraman/RoBERTa-TR-medium-char

RoBERTa Turkish medium Character-level (uncased)

Pretrained model on Turkish language using a masked language modeling (MLM) objective. The model is uncased. The pretrained corpus is OSCAR's Turkish split, but it is further filtered and cleaned.

Model architecture is similar to bert-medium (8 layers, 8 heads, and 512 hidden size). Tokenization algorithm is Character-level, which means that text is split by individual characters. Vocabulary size is 384.

The details and performance comparisons can be found at this paper: https://arxiv.org/abs/2204.08832

Note that this model does not include a tokenizer file, because it uses ByT5Tokenizer. The following code can be used for model loading and tokenization, example max length(1024) can be changed:

    model = AutoModel.from_pretrained([model_path])
    #for sequence classification:
    #model = AutoModelForSequenceClassification.from_pretrained([model_path], num_labels=[num_classes])

    tokenizer = ByT5Tokenizer.from_pretrained("google/byt5-small")
    tokenizer.mask_token = tokenizer.special_tokens_map_extended['additional_special_tokens'][0]
    tokenizer.cls_token = tokenizer.special_tokens_map_extended['additional_special_tokens'][1]
    tokenizer.bos_token = tokenizer.special_tokens_map_extended['additional_special_tokens'][1]
    tokenizer.sep_token = tokenizer.special_tokens_map_extended['additional_special_tokens'][2]
    tokenizer.eos_token = tokenizer.special_tokens_map_extended['additional_special_tokens'][2]
    tokenizer.pad_token = tokenizer.special_tokens_map_extended['additional_special_tokens'][3]
    tokenizer.unk_token = tokenizer.special_tokens_map_extended['additional_special_tokens'][3]
    tokenizer.model_max_length = 1024

BibTeX entry and citation info

@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}
}