Mahan Nano Base

This repository is for the Mahan Nano base model. The current artifact is a tokenizer; no trained model weights are supplied.

Tokenizer files and the training report are in tokenizer/.

from transformers import AutoTokenizer
t = AutoTokenizer.from_pretrained('mahan13900/Mahan-Nano-Base', subfolder='tokenizer', token=True)
ids = t('سلام می‌روم', add_special_tokens=False, split_special_tokens=True)['input_ids']
ids.append(t.eos_token_id)

Special tokens alone do not create reasoning or fill-in-the-middle ability. Evaluate downstream behavior after training the model.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support