This is a tokenizer with same settings as bert-base-uncased but sets strip_accents=False.
Has a vocab size of 30K and was trained on various corpus including Nepali wikipedia, new articles etc.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("jangedoo/ne_bert_tiny", dtype="auto")