Autoencoding Transformers
mT5
Introduced by Xue et al. in mT5: A massively multilingual pre-trained text-to-text transformermt5 is a multilingual variant of T5 that was pre-trained on a new Common Crawl-based dataset covering $101$ languages.
Source: mT5: A massively multilingual pre-trained text-to-text transformer
Papers
| Paper | Code | Results | Date | Stars |
|---|
Tasks
| Task | Papers | Share |
|---|---|---|
| Language Modelling | 12 | 10.71% |
| Question Answering | 10 | 8.93% |
| Machine Translation | 8 | 7.14% |
| Abstractive Text Summarization | 7 | 6.25% |
| XLM-R | 5 | 4.46% |
| Denoising | 4 | 3.57% |
| Cross-Lingual Transfer | 4 | 3.57% |
| Text Generation | 4 | 3.57% |
| Natural Language Understanding | 4 | 3.57% |
T5
