CoDAR: Continuous Diffusion Language Models are More Powerful Than You Think
This work addresses the limitations of continuous diffusion language models for natural language processing tasks, providing an incremental yet significant improvement for the research community.
The authors tackled the bottleneck of continuous diffusion language models, achieving a substantial improvement in generation quality, making them competitive with strong discrete models. CoDAR outperformed latent diffusion and exposed a decoder-temperature knob to navigate the fluency-diversity trade-off.
We study why continuous diffusion language models (DLMs) have lagged behind discrete diffusion approaches despite their appealing continuous generative dynamics. Under a controlled token--recovery study, we identify token rounding, the final projection from denoised embeddings to tokens, as a primary bottleneck. Building on these insights, we propose CoDAR (Continuous Diffusion with Contextual AutoRegressive Decoder), a two--stage framework that keeps diffusion entirely continuous in an embedding space while learning a strong, context--conditional discretizer: an autoregressive Transformer decoder that cross--attends to the denoised embedding sequence and performs contextualized rounding to tokens. Experiments on LM1B and OpenWebText demonstrate that CoDAR substantially improves generation quality over latent diffusion and becomes competitive with strong discrete DLMs, while exposing a simple decoder--temperature knob to navigate the fluency--diversity trade off.