English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Diffusion models have been successfully adapted to text generation tasks bymapping the discrete text into the continuous space. However, there existnonnegligible gaps between training and inference, owing to the absence of theforward process during inference. Thus, the model only predicts based on thepreviously generated reverse noise rather than the noise computed by theforward process. Besides, the widely-used downsampling strategy in speeding upthe inference will cause the mismatch of diffusion trajectories betweentraining and inference. To understand and mitigate the above two types oftraining-inference discrepancies, we launch a thorough preliminary study. Basedon our observations, we propose two simple yet effective methods to bridge thegaps mentioned above, named Distance Penalty and Adaptive Decay Sampling.Extensive experiments on \textbf{6} generation tasks confirm the superiority ofour methods, which can achieve $100\times \rightarrow 200\times$ speedup withbetter performance.

Can Diffusion Model Achieve Better Performance in Text Generation? Bridging the Gap between Training and Inference!