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 to our paper
ELECTRA: Pre-training Text Encoders as Discriminators Rather Than Generators.
This repository contains code to pre-train ELECTRA, including small ELECTRA models on a single GPU. It also supports fine-tuning ELECTRA on downstream tasks including classification tasks (e.g,.
GLUE), QA tasks (e.g.,
SQuAD), and sequence tagging tasks (e.g.,
text chunking).
1from transformers import ElectraForPreTraining, ElectraTokenizerFast
2import torch
3
4discriminator = ElectraForPreTraining.from_pretrained("google/electra-base-discriminator")
5tokenizer = ElectraTokenizerFast.from_pretrained("google/electra-base-discriminator")
6
7sentence = "The quick brown fox jumps over the lazy dog"
8fake_sentence = "The quick brown fox fake over the lazy dog"
9
10fake_tokens = tokenizer.tokenize(fake_sentence)
11fake_inputs = tokenizer.encode(fake_sentence, return_tensors="pt")
12discriminator_outputs = discriminator(fake_inputs)
13predictions = torch.round((torch.sign(discriminator_outputs[0]) + 1) / 2)
14
15[print("%7s" % token, end="") for token in fake_tokens]
16
17[print("%7s" % int(prediction), end="") for prediction in predictions.tolist()]
