Zero-Shot Translation using Diffusion Models
This addresses the problem of translation efficiency and generalization for NLP researchers, though it appears incremental as it adapts existing diffusion models to text.
The authors tackled neural machine translation by proposing a novel non-autoregressive method using a denoising diffusion probabilistic model conditioned on source sentences, achieving zero-shot translation between language pairs unseen during training.
In this work, we show a novel method for neural machine translation (NMT), using a denoising diffusion probabilistic model (DDPM), adjusted for textual data, following recent advances in the field. We show that it's possible to translate sentences non-autoregressively using a diffusion model conditioned on the source sentence. We also show that our model is able to translate between pairs of languages unseen during training (zero-shot learning).