Metadata-Version: 2.1
Name: biobert-pytorch
Version: 0.9
Summary: Ready to use BioBert pytorch weights for HuggingFace pytorch BertModel.
Home-page: https://github.com/jgamper/biobert-pytorch
Author: Jevgenij Gamper
Author-email: jevgenij.gamper5@gmail.com
License: UNKNOWN
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3.6
Description-Content-Type: text/markdown

# Biobertology

Ready to use [BioBert](https://arxiv.org/abs/1901.08746) pytorch weights for [HuggingFace](https://github.com/huggingface/transformers)
pytorch [BertModel](https://arxiv.org/abs/1810.04805).

To load the model:

```
from biobertology import get_biobert, get_tokenizer

biobert = get_biobert(model_dir=None, download=True)
tokenizer = get_tokenizer()
```

Example of fine tuning biobert [here](https://github.com/MeRajat/SolvingAlmostAnythingWithBert/tree/master/biobert_ner).

## How was it converted to pytorch?

Model weights have been downloaded from [here](https://github.com/naver/biobert-pretrained/releases/tag/v1.1-pubmed) and converted

by following the commands described [here](https://github.com/huggingface/transformers/issues/457#issuecomment-518403170).pytorch

Specifically, using the command:

```
pytorch_transformers bert biobert_v1.1_pubmed/model.ckpt-1000000 biobert_v1.1_pubmed/bert_config.json biobert_v1.1_pubmed/pytorch_model.bin
```

followed by the renaming of the config file:

```
mv biobert_v1.1_pubmed/bert_config.json biobert_v1.1_pubmed/config.json
```

all of the other files inside the original weights file have been deleted.

