hankcs--hanlp/docs/api/hanlp/pretrained/mlm.md

jupytext, kernelspec

jupytext	kernelspec
formats text_representation ipynb,md:myst extension format_name format_version jupytext_version .md myst 0.8 1.4.2	display_name language name Python 3 python python3

mlm

Masked Language Model (MLM) predicts words that were originally hidden intentionally in a sentence. To perform such prediction, first load a pre-trained MLM (e.g., bert-base-chinese):

```{hint}
Predict multiple sentences in batch mode for faster speed! 
```

```{note}
HanLP always support multilingual. Feel free to use a multilingual model listed [here](https://huggingface.co/models?pipeline_tag=fill-mask&sort=downloads).
```

:tags: [output_scroll]
from hanlp.components.lm.mlm import MaskedLanguageModel
mlm = MaskedLanguageModel()
mlm.load('bert-base-chinese')

Represent blanks (masked tokens) with [MASK] and let MLM fills them:

:tags: [output_scroll]
mlm('生活的真谛是[MASK]。')

Batching is always faster:

:tags: [output_scroll]
mlm(['生活的真谛是[MASK]。', '巴黎是[MASK][MASK]的首都。'])

All the pre-trained MLM models and their details are listed in the docs of Hugging Face 🤗 Transformers.