Transformers
RoBERTa
Introduced by Liu et al. in RoBERTa: A Robustly Optimized BERT Pretraining ApproachRoBERTa is an extension of BERT with changes to the pretraining procedure. The modifications include:
- training the model longer, with bigger batches, over more data
- removing the next sentence prediction objective
- training on longer sequences
- dynamically changing the masking pattern applied to the training data. The authors also collect a large new dataset ($\text{CC-News}$) of comparable size to other privately used datasets, to better control for training set size effects
Papers
| Paper | Code | Results | Date | Stars |
|---|
Tasks
| Task | Papers | Share |
|---|---|---|
| Language Modelling | 72 | 8.25% |
| Sentence | 49 | 5.61% |
| Sentiment Analysis | 44 | 5.04% |
| Question Answering | 37 | 4.24% |
| Text Classification | 36 | 4.12% |
| Classification | 22 | 2.52% |
| NER | 19 | 2.18% |
| parameter-efficient fine-tuning | 17 | 1.95% |
| Decoder | 16 | 1.83% |
BERT
