|
--- |
|
language: it |
|
license: mit |
|
datasets: |
|
- wikipedia |
|
--- |
|
|
|
# 🤗 + 📚 dbmdz BERT and ELECTRA models |
|
|
|
In this repository the MDZ Digital Library team (dbmdz) at the Bavarian State |
|
Library open sources Italian BERT and ELECTRA models 🎉 |
|
|
|
# Italian BERT |
|
|
|
The source data for the Italian BERT model consists of a recent Wikipedia dump and |
|
various texts from the [OPUS corpora](http://opus.nlpl.eu/) collection. The final |
|
training corpus has a size of 13GB and 2,050,057,573 tokens. |
|
|
|
For sentence splitting, we use NLTK (faster compared to spacy). |
|
Our cased and uncased models are training with an initial sequence length of 512 |
|
subwords for ~2-3M steps. |
|
|
|
For the XXL Italian models, we use the same training data from OPUS and extend |
|
it with data from the Italian part of the [OSCAR corpus](https://traces1.inria.fr/oscar/). |
|
Thus, the final training corpus has a size of 81GB and 13,138,379,147 tokens. |
|
|
|
Note: Unfortunately, a wrong vocab size was used when training the XXL models. |
|
This explains the mismatch of the "real" vocab size of 31102, compared to the |
|
vocab size specified in `config.json`. However, the model is working and all |
|
evaluations were done under those circumstances. |
|
See [this issue](https://github.com/dbmdz/berts/issues/7) for more information. |
|
|
|
The Italian ELECTRA model was trained on the "XXL" corpus for 1M steps in total using a batch |
|
size of 128. We pretty much following the ELECTRA training procedure as used for |
|
[BERTurk](https://github.com/stefan-it/turkish-bert/tree/master/electra). |
|
|
|
## Model weights |
|
|
|
Currently only PyTorch-[Transformers](https://github.com/huggingface/transformers) |
|
compatible weights are available. If you need access to TensorFlow checkpoints, |
|
please raise an issue! |
|
|
|
| Model | Downloads |
|
| ---------------------------------------------------- | --------------------------------------------------------------------------------------------------------------- |
|
| `dbmdz/bert-base-italian-cased` | [`config.json`](https://cdn.huggingface.co/dbmdz/bert-base-italian-cased/config.json) • [`pytorch_model.bin`](https://cdn.huggingface.co/dbmdz/bert-base-italian-cased/pytorch_model.bin) • [`vocab.txt`](https://cdn.huggingface.co/dbmdz/bert-base-italian-cased/vocab.txt) |
|
| `dbmdz/bert-base-italian-uncased` | [`config.json`](https://cdn.huggingface.co/dbmdz/bert-base-italian-uncased/config.json) • [`pytorch_model.bin`](https://cdn.huggingface.co/dbmdz/bert-base-italian-uncased/pytorch_model.bin) • [`vocab.txt`](https://cdn.huggingface.co/dbmdz/bert-base-italian-uncased/vocab.txt) |
|
| `dbmdz/bert-base-italian-xxl-cased` | [`config.json`](https://cdn.huggingface.co/dbmdz/bert-base-italian-xxl-cased/config.json) • [`pytorch_model.bin`](https://cdn.huggingface.co/dbmdz/bert-base-italian-xxl-cased/pytorch_model.bin) • [`vocab.txt`](https://cdn.huggingface.co/dbmdz/bert-base-italian-xxl-cased/vocab.txt) |
|
| `dbmdz/bert-base-italian-xxl-uncased` | [`config.json`](https://cdn.huggingface.co/dbmdz/bert-base-italian-xxl-uncased/config.json) • [`pytorch_model.bin`](https://cdn.huggingface.co/dbmdz/bert-base-italian-xxl-uncased/pytorch_model.bin) • [`vocab.txt`](https://cdn.huggingface.co/dbmdz/bert-base-italian-xxl-uncased/vocab.txt) |
|
| `dbmdz/electra-base-italian-xxl-cased-discriminator` | [`config.json`](https://s3.amazonaws.com/models.huggingface.co/bert/dbmdz/electra-base-italian-xxl-cased-discriminator/config.json) • [`pytorch_model.bin`](https://cdn.huggingface.co/dbmdz/electra-base-italian-xxl-cased-discriminator/pytorch_model.bin) • [`vocab.txt`](https://cdn.huggingface.co/dbmdz/electra-base-italian-xxl-cased-discriminator/vocab.txt) |
|
| `dbmdz/electra-base-italian-xxl-cased-generator` | [`config.json`](https://s3.amazonaws.com/models.huggingface.co/bert/dbmdz/electra-base-italian-xxl-cased-generator/config.json) • [`pytorch_model.bin`](https://cdn.huggingface.co/dbmdz/electra-base-italian-xxl-cased-generator/pytorch_model.bin) • [`vocab.txt`](https://cdn.huggingface.co/dbmdz/electra-base-italian-xxl-cased-generator/vocab.txt) |
|
|
|
## Results |
|
|
|
For results on downstream tasks like NER or PoS tagging, please refer to |
|
[this repository](https://github.com/stefan-it/italian-bertelectra). |
|
|
|
## Usage |
|
|
|
With Transformers >= 2.3 our Italian BERT models can be loaded like: |
|
|
|
```python |
|
from transformers import AutoModel, AutoTokenizer |
|
|
|
model_name = "dbmdz/bert-base-italian-cased" |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name) |
|
|
|
model = AutoModel.from_pretrained(model_name) |
|
``` |
|
|
|
To load the (recommended) Italian XXL BERT models, just use: |
|
|
|
```python |
|
from transformers import AutoModel, AutoTokenizer |
|
|
|
model_name = "dbmdz/bert-base-italian-xxl-cased" |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name) |
|
|
|
model = AutoModel.from_pretrained(model_name) |
|
``` |
|
|
|
To load the Italian XXL ELECTRA model (discriminator), just use: |
|
|
|
```python |
|
from transformers import AutoModel, AutoTokenizer |
|
|
|
model_name = "dbmdz/electra-base-italian-xxl-cased-discriminator" |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name) |
|
|
|
model = AutoModelWithLMHead.from_pretrained(model_name) |
|
``` |
|
|
|
# Huggingface model hub |
|
|
|
All models are available on the [Huggingface model hub](https://huggingface.co./dbmdz). |
|
|
|
# Contact (Bugs, Feedback, Contribution and more) |
|
|
|
For questions about our BERT/ELECTRA models just open an issue |
|
[here](https://github.com/dbmdz/berts/issues/new) 🤗 |
|
|
|
# Acknowledgments |
|
|
|
Research supported with Cloud TPUs from Google's TensorFlow Research Cloud (TFRC). |
|
Thanks for providing access to the TFRC ❤️ |
|
|
|
Thanks to the generous support from the [Hugging Face](https://huggingface.co./) team, |
|
it is possible to download both cased and uncased models from their S3 storage 🤗 |
|
|
