• XLNet

  • Referenced in 7 articles [sw31118]
  • formulation. Furthermore, XLNet integrates ideas from Transformer-XL, the state-of-the-art autoregressive model...
  • Transformer-XL

  • Referenced in 3 articles [sw36208]
  • Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context. Transformers have a potential ... propose a novel neural architecture Transformer-XL that enables learning dependency beyond a fixed length ... context fragmentation problem. As a result, Transformer-XL learns dependency that is 80% longer than ... When trained only on WikiText-103, Transformer-XL manages to generate reasonably coherent, novel text...
  • TensorFlow

  • Referenced in 363 articles [sw15170]
  • TensorFlow™ is an open source software library for...
  • PyTorch

  • Referenced in 198 articles [sw20939]
  • PyTorch python package: Tensors and Dynamic neural networks...
  • GLUE

  • Referenced in 5 articles [sw30755]
  • The General Language Understanding Evaluation (GLUE) benchmark is...
  • BERT

  • Referenced in 38 articles [sw30756]
  • BERT: Pre-training of Deep Bidirectional Transformers for...
  • MaskGAN

  • Referenced in 3 articles [sw31828]
  • MaskGAN: Better Text Generation via Filling in the...
  • RoBERTa

  • Referenced in 8 articles [sw32571]
  • RoBERTa: A Robustly Optimized BERT Pretraining Approach. RoBERTa...
  • SentencePiece

  • Referenced in 2 articles [sw35795]
  • SentencePiece: A simple and language independent subword tokenizer...
  • ALBERT

  • Referenced in 3 articles [sw36207]
  • ALBERT: A Lite BERT for Self-supervised Learning...
  • MADE

  • Referenced in 9 articles [sw36209]
  • MADE: Masked Autoencoder for Distribution Estimation. There has...
  • TopicRNN

  • Referenced in 3 articles [sw36211]
  • TopicRNN: A Recurrent Neural Network with Long-Range...
  • cmix

  • Referenced in 1 article [sw36212]
  • cmix is a lossless data compression program aimed...
  • DARTS

  • Referenced in 7 articles [sw36213]
  • DARTS: Differentiable Architecture Search. This paper addresses the...