mrm8488 / electricidad-base-discriminator

ELECTRICIDAD: The Spanish Electra Imgur

Electricidad-base-discriminator (uncased) is a base Electra like model (discriminator in this case) trained on a Large Spanish Corpus (aka BETO's corpus)

As mentioned in the original paper : ELECTRA is a new method for self-supervised language representation learning. It can be used to pre-train transformer networks using relatively little compute. ELECTRA models are trained to distinguish "real" input tokens vs "fake" input tokens generated by another neural network, similar to the discriminator of a GAN . At small scale, ELECTRA achieves strong results even when trained on a single GPU. At large scale, ELECTRA achieves state-of-the-art results on the SQuAD 2.0 dataset.

For a detailed description and experimental results, please refer the paper ELECTRA: Pre-training Text Encoders as Discriminators Rather Than Generators .

Model details ⚙

Evaluation metrics (for discriminator) 🧾

Fast example of usage 🚀

from transformers import ElectraForPreTraining, ElectraTokenizerFast
import torch

discriminator = ElectraForPreTraining.from_pretrained("mrm8488/electricidad-base-discriminator")
tokenizer = ElectraTokenizerFast.from_pretrained("mrm8488/electricidad-base-discriminator")

sentence = "El rápido zorro marrón salta sobre el perro perezoso"
fake_sentence = "El rápido zorro marrón amar sobre el perro perezoso"

fake_tokens = tokenizer.tokenize(fake_sentence)
fake_inputs = tokenizer.encode(fake_sentence, return_tensors="pt")
discriminator_outputs = discriminator(fake_inputs)
predictions = torch.round((torch.sign(discriminator_outputs[0]) + 1) / 2)

[print("%7s" % token, end="") for token in fake_tokens]

[print("%7s" % prediction, end="") for prediction in predictions.tolist()]

# Output:
'''
el rapido  zorro  marro    ##n   amar  sobre     el  perro   pere ##zoso    0.0    0.0    0.0    0.0    0.0    0.0    1.0    1.0    0.0    0.0    0.0    0.0    0.0[None, None, None, None, None, None, None, None, None, None, None, None, None
'''

As you can see there are 1s in the places where the model detected a fake token. So, it works! 🎉

Some models fine-tuned on a downstream task 🛠️

Question Answering

POS

NER

Spanish LM model comparison 📊

Acknowledgments

I thank 🤗/transformers team for allowing me to train the model (specially to Julien Chaumond ).

Citation

If you want to cite this model you can use this:

@misc{mromero2020electricidad-base-discriminator,
  title={Spanish Electra by Manuel Romero},
  author={Romero, Manuel},
  publisher={Hugging Face},
  journal={Hugging Face Hub},
  howpublished={\url{https://huggingface.co/mrm8488/electricidad-base-discriminator/}},
  year={2020}
}

Created by Manuel Romero/@mrm8488

Made with ♥ in Spain