New Multilingual Encoder-Decoder Models for Seq2Seq Tasks
Hi everyone, we’re excited to announce a new collection of multilingual encoder-decoder models for sequence-to-sequence tasks. Here are some of the models we’ve curated:
These models have been reported to achieve state-of-the-art NLI scores, and we’re looking forward to seeing how they perform in multilingual settings. We’re also keeping an eye on open questions, such as using mamba for encoder-only architectures. Stay tuned for more updates!
For more information, check out the Transformers documentation and the mamba GitHub issue.