README.md

---

language: 
- tr
tags:
- roberta
license: cc-by-nc-sa-4.0
datasets:
- oscar
---


# RoBERTa Turkish medium Morph-level 7k (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 Morph-level, which means that text is split according to a Turkish morphological analyzer (Zemberek). Vocabulary size is 7.5k. 

## Note that this model needs a preprocessing step before running, because the tokenizer file is not a morphological anaylzer. That is, the test dataset can not be split into morphemes with the tokenizer file. The user needs to process any test dataset by a Turkish morphological analyzer (Zemberek in this case) before running evaluation.

The details can be found at this paper: 
https://arxiv.org/...

The following code can be used for model loading and tokenization, example max length (514) can be changed:
```

	model = AutoModel.from_pretrained([model_path])

	#for sequence classification:

	#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 entry and citation info
```bibtex

@article{}

```