elbayadm/attn2d
A neural machine translation toolkit extending Fairseq with 2D convolutional networks for sequence-to-sequence prediction.
Not currently ranked — collecting fresh signals.
star history
This repository provides implementations of Pervasive Attention, which uses 2D convolutional neural networks to jointly encode source and target sequences for neural machine translation. It also includes efficient Wait-k models for simultaneous machine translation. Built as a fork of Fairseq, it leverages PyTorch for deep learning operations and supports training wait-k models with joint training of multiple wait-k paths.
Frequently asked
- What is elbayadm/attn2d?
- A neural machine translation toolkit extending Fairseq with 2D convolutional networks for sequence-to-sequence prediction.
- Is attn2d open source?
- Yes — elbayadm/attn2d is open source, released under the MIT license.
- What language is attn2d written in?
- elbayadm/attn2d is primarily written in Python.
- How popular is attn2d?
- elbayadm/attn2d has 497 stars on GitHub.
- Where can I find attn2d?
- elbayadm/attn2d is on GitHub at https://github.com/elbayadm/attn2d.