frontier models
A Continuous Latent Kept Parallel Tokens Connected
HC-DLM improved structured reasoning and perplexity at matched model size.
Published Updated Story ID: mp-2026-10-02-016
Summary
HC-DLM improved structured reasoning and perplexity at matched model size.
Hierarchical Continuous Diffusion Language Models use one persistent latent trajectory while decoded tokens feed back as a scaffold. On Sudoku, Countdown and LM1B, the reported model beat discrete and continuous diffusion baselines in puzzle accuracy and generative perplexity.
Why it matters
HC-DLM improved structured reasoning and perplexity at matched model size.
Limits and context
No additional limitation was separately recorded.
Key claims
HC-DLM improved structured reasoning and perplexity at matched model size.
Evidence: source-2026-10-02-018
Sources
- arXiv preprint 2610.02193arXiv · primary research
Corrections
No corrections have been recorded for this story.