H-CDLM brings DINO-style hierarchical semantics to continuous diffusion LMs for cheaper generation
sedielem · x · 2026-10-19
New paper introduces Hierarchical Continuous Diffusion LMs (H-CDLMs), which jointly diffuse tokens and coarse semantic clusters built via clustering. Sedielem notes DINO-style semantic components already play a big role in visual diffusion models, and this work ports the idea to continuous diffusion LMs—yielding better generations for less compute.
More from Research
- Reading activations changes compiled Qwen3's logits in up to 93% of tokens — Sauers_ · 2026-10-19
- Oxford DPhil thesis argues Attribution-Based Control could fix hallucination, alignment and power concentration at once — iamtrask · 2026-10-19
- Stanford's Memorilla: 16 memory tokens beat full-context with 108x fewer tokens — omarsar0 · 2026-10-19
- Philosopher's "AI and Consciousness: A Skeptical Overview" Free Online This Week — eschwitz · 2026-10-19
- Handpicked 40-Game Puzzlescript Benchmark Aims to Be "ARC-AGI-3.5" for Agent Reasoning — scaling01 · 2026-10-19
- LeCun: distributed representations beat grandmother cells, known since the 1980s — burny_tech · 2026-10-19